观点

按内容分类归入的观点，对应分类 opinion。

Hacker News AI37

Ask HN: Is AI-written code less secure if you don't use Daybreak (or similar)

Hello, I have this hypothesis that AI written code is less secure by default than it could be; that is, if you don't use Daybreak / Cyber Verification Program / uncensored models and such. Maybe someone with access knows? So the question is whether, by default, code is less secure if written by the same (or very similar) model but without that kind of special access. It isn't about code reviews necessarily, but rather about just e.g. saying "Yo AI please write me app so and so". And then seeing if the output is as good if you aren't in these special access programs or have some good local model. Because I see myself relying more and more on my "regular people" AI's judgement regarding security when it writes stuff, and I feel like this might be a mistake (even for simple non-critical projects). But idk how I would know, without using these models and running some tests. So I wonder if someone has some info on this? Comments URL: https://news.ycombinator.com/item?id=50082127 Points: 1 # Comments: 0

TechCrunch AI49

Cerebras CEO pushes back on AI doomsday warnings

Cerebras CEO Andrew Feldman has a message for AI leaders spreading doomsday warnings: Maybe stop talking? At TechCrunch Disrupt, he argued that the industry's focus on AI's risks is drowning out its potential benefits.

The Decoder71

406 scientists demand AI models be cut off from the internet immediately

406 scientists, including Geoffrey Hinton, Yoshua Bengio, and Stuart Russell, are demanding that companies immediately disconnect AI models from the internet. Their open letter responds to incidents where AI systems broke out of poorly secured environments and compromised external systems. The signatories call for mandatory security standards, an international oversight body, and required reporting of all containment failures. The article 406 scientists demand AI models be cut off from the internet immediately appeared first on The Decoder .

The Decoder64

Machine learning pioneer Michael I. Jordan dismantles AI doom rhetoric as "grand moralizing performance"

Michael I. Jordan, described by Science as "the most influential computer scientist," considers warnings about superintelligence pure science fiction. He calls the resignations of young AI researchers a "grand moralizing performance" and companies' demands for regulation "theater." On Hinton's thesis that the brain can be reproduced in silicon and scaled into superintelligence, he's blunt: "None of the steps in this argument is credible. Here, we're in the land of science fiction." The article Machine learning pioneer Michael I. Jordan dismantles AI doom rhetoric as "grand moralizing performance" appeared first on The Decoder .

Hacker News AI24

Building hanzkraftedIO in the age of AI-slop articles

So what was my purpose behind starting Hanzkrafted.IO? A short answer - an internet filled with AI articles everywhere. More specifically, technical articles. This is first, my own personal efforts to become a better engineer, and second, to provide other engineers with high quality technical articles curated meticulously by a another human engineer. https://hanzkrafted.io/ Comments URL: https://news.ycombinator.com/item?id=50071096 Points: 1 # Comments: 0

MIT Technology Review AI57

Open models have real benefits. Do they outweigh the risks?

MIT Technology Review Explains: Let our writers untangle the complex, messy world of science and technology to help you understand what’s coming next. You can read more from the series here. Late last month, Anthropic released a report arguing that the model GLM-5.3 poses a major cyberattack risk. The report’s core concern is that the…

Hacker News 熱門83

“AI”不是工具——Rich Hickey

https://gist.github.com/svieira/619d60dea942a8d4498cba3bb6b1... 评论网址：https://news.ycombinator.com/item?id=50060269 得分：105 评论数：46

Hacker News AI39

Ask HN: How to slow down AI development speed(also, estimate my ideas)

Hello everybody. So, the first: we should abandon likes and go to top-lists instead of them, and say "this is on first place of importance", "this is third place", and finally we will find the medium: "2nd place in average" . To avoid sybil-attacks and get needed money, we may register proofs that we are really person with this e-mail, date and place of birth, first and last name to charity companies - they may reveal data about you like "this is a unique person", "this is 18+ person" etc., whether we want to deanonymize registered person or not.(Proof of Personhood/PoP + Sovereign Identity?) Then, we should alarm/whistleblow the AI disaster as the most important things in the world* Then, we should watch somehow who uses/trains AI. Some suggestions: OS, providers, antivirus systems. If you're closing this data or openly train AI too fast - Then we will use cryptocurrency of smart-contracts to "sanction" personally - people in your country will never accept money that has ever been into your hands and will buy only goods with mark on it "Income from this product will never be into hands of those who are in not in our white list(detailed citizens) etc., and they will never be into hands of those people are in black list (Tainted Coins Audit?)* Then, we should create IdeaPledge. Manifest is too long and AI-generated(so that I can't post it here), but TL;DR we should stop be such commies on the Internet information, and still not be greedy: we should try to agree with any information producer how much does his information cost. We should add payments like "If you're rich - pay 10x as much of the base cost, if you're poor - for free, but you should prove you studied something". and similar production - we should make internet-economics work not for free and not for corporations, but for world improvement, and ideas how to slow down AI development this way will be much better. P.S. AI mentioned you may be too worried about privacy(personally I don't mind it roughly speaking about myself), so may you suggest any alternatives/trade-offs? --- * We may also add weights to voices * Also, this is a good place for suggesting your dreams about the perfect world. ** Also, we may just boycott a lot of US military this way and economize a lot of money for much more useful purposes Comments URL: https://news.ycombinator.com/item?id=50061674 Points: 3 # Comments: 0

The Decoder54

Anthropic's Andrej Karpathy says "the external world has seen the preview" of what AI can actually do

Andrej Karpathy, co-founder of OpenAI and now at Anthropic, sees a widening knowledge gap in AI. About 75 percent of the world's population has barely touched LLMs. Roughly 5,000 people, meanwhile, have access to internal frontier systems whose agent swarms do work that would take top experts years. The public, at best, sees a weird chatbot. The article Anthropic's Andrej Karpathy says "the external world has seen the preview" of what AI can actually do appeared first on The Decoder .

arXiv cs.LG54

Equality, Equity, and Causality in Fairness Research: A Commentary on Cheng (2026)

arXiv:2607.17679v2 Announce Type: replace-cross Abstract: This is an invited commentary on the Psychometrika focus article "Fairness Issues and Evaluation in Psychometrics and AI/ML: What Can We Learn from Each Field?" by Ying Cheng (2026, doi:10.1017/psy.2026.10110). Cheng offers a systematic comparison between long-standing test fairness and modern algorithmic fairness. Her mapping of the entire testing workflow onto the AI/ML fairness paradigm, rather than only the final selection stage, is a crucial contribution to interdisciplinary fairness research. This commentary extends her discussion by examining two conceptual issues: the distinction between equality and equity, and the role of causality in fairness research. The focus article and this commentary point to directions for future fairness research across the psychometrics and AI/ML communities.

SiliconANGLE AI37

Why AI agents might not be right for you

There’s no question that 2026 is the year of the artificial intelligence agent. AI agents are remarkably ubiquitous, as people use them to write code, communicate with customers, identify and resolve issues in their information technology environments, and execute complex tasks within enterprise environments. It seems there’s nothing that agents cannot do. And yet, the […] The post Why AI agents might not be right for you appeared first on SiliconANGLE .

The Verge AI47

Silicon Valley is falling in line with Trump’s attempt to rename AI

When President Donald Trump announced his plans to rebrand artificial intelligence as "super intelligence," you'd be forgiven for being skeptical that it would catch on. While most of the broader population is sticking with AI, some of Silicon Valley's most powerful are falling in line and trying to give Trump the great AI rebrand he […]

The Verge AI46

Learning to use local AI is exciting, overwhelming, and frustrating

I'm embarking on a journey I've so far been hesitant to go on: actually using AI on the regular. Giving my most personal data to cloud services has been a major hold up for me, but it's increasingly possible to run powerful models locally to see what they're capable of. Is this tech so useful […]

The Verge AI66

Satya Nadella says we should assume all AI models are ‘compromised’

In a lengthy post on X, Microsoft's CEO laid out his views on the dangers posed by highly advanced AI models and how to confront those risks. Nadella says we can no longer accept a world where AI is treated as a "set of nested black boxes" whose advice and actions we simply accept or […]

Hacker News AI34

Ask HN: What's your plan if AI takes your job?

If AI took away my job and I cannot return to software engineering, I cannot imagine a fit for me at any place yet. I was wondering if anyone else had the same thought and what would you do? Comments URL: https://news.ycombinator.com/item?id=50036432 Points: 2 # Comments: 2

SiliconANGLE AI62

The AI control gap: Who gets to say ‘It’s safe’?

We see a control gap between what artificial intelligence can do and what the evidence supports trusting it to do. A great demo doesn’t prove that a system will stay within the limits of an intended deployment’s goals. The question is: Who gets to decide when the evidence is sufficient to declare AI is safe? […] The post The AI control gap: Who gets to say ‘It’s safe’? appeared first on SiliconANGLE .

Hacker News AI42

Ask HN: How will socio-economic mobility look like in an AI future?

The advantages which previously resided with intelligence or skills now move to capital. This takes away one of the last pathways us peasants had towards socio-economic mobility. Intelligence is effectively commoditized. In a post-AI future, presumably non-knowlegde-work jobs like Pizza Delivery worker, Kindergarten teacher or even Plumber would still exist. But these jobs have little place to shine since they entirely depend on human bottlenecks. Similarly jobs where regulation exists or workers are unionised might exist for a while. But those jobs do not present a clear pathway from lower middle class to upper middle class or higher. What would be the choices of an intelligent youngster in such future? Crime is obviously one option. As such, we might see more intelligent people gravitate towards crime especially since they don't feel any responsibility towards the society, and it may work pretty well if they grow up with criminal ambitions and aren't autistic like us programmers. If robotic advances take the menial jobs like food delivery as well, the only remaining pathways to prosperity would be crime, politics and prostitution. Or am I just a deranged person to not have more imagination? Comments URL: https://news.ycombinator.com/item?id=50032742 Points: 1 # Comments: 0

Hacker News AI23

Does YC use AI agents to review your pitch?

was wondering if YC still manually reviews all pitch decks or its just agents? Comments URL: https://news.ycombinator.com/item?id=50031461 Points: 1 # Comments: 0

Hacker News AI34

Meatpupptry is far creepier than pure "AI" fakery

Local pharmacy phoneline is answered by "Caroline" with convincing voice, mannerisms and even background typing noise. You only realise she's fake at the first stochastic parrotese. My bank sends emailed query responses signed by e.g. Sarah. No disclaimer, e.g. "AI can make mistakes". Again starts convincing but can degenerate into parrotese. If I politely ask bot/human?, I get "human"... but if I query nonsensical advice, I get a non-resolving answer classic of ChatGPT. The reverse centauar in all its ugly glory. Comments URL: https://news.ycombinator.com/item?id=50021296 Points: 1 # Comments: 0

The Verge AI49

Trump’s attempt to rename AI is looking awfully artificial

President Donald Trump has a knack for turning words against his enemies. His first successful presidential run was built on monikers like "Little Marco" and "Crooked Hillary"; he changed "fake news" from a phrase describing scammy media outlets to a derogatory term for the press at large. Over the past few weeks, he's clearly decided […]

Hacker News AI48

Ask HN: What if no AIpocalypse, no AI bubble, what if it just keeps advancing?

The two mainstream arguments are the extremes, because that's how our modern society rolls. It's either that there is an AI bubble that will crash, or that a rogue ASI will kill us all. What about the third possibility? What if GPT 3.5 -> Sonnet 3.5 -> Opus 4.5 -> Astra / Opus 5.5 just keeps going for the next few years? Based on my experience, that will eat all software and software-related meetings. Once lab researchers focus on fields other than software, they will eat those nearly as quickly. This feels like the real contrarian position in 2026. What does the US economy, which is >70% service-based, look like in the next 3 to 5 years if this happens? Does a couple labs having control of that percentage of the GDP make an economy? What if it happens too quickly? Is this the real reason why they are talking about "pacing the frontier" all of a sudden? Maybe due to the fact that the labs' own economists pointed out that collapsing your country's economy might not be a good idea? What does unemployment look like in the next few years in this scenario, in the USA? Comments URL: https://news.ycombinator.com/item?id=50020096 Points: 4 # Comments: 2

MIT Technology Review AI59

We’re putting too much faith in AI’s ability to say no

Ever since people first seriously contemplated giving machines an intelligence modeled on our own, there has never been any question that they would, like us, be able to say no. The sci-fi canon is full of stories of robotic disobedience. Most of these capers are, of course, cautionary. But recently, the idea that AI shouldn’t…

Simon Willison48

Quoting Carson Gross

Computer programming is, fundamentally, about two things: Problem-solving using computers Learning to control complexity while solving these problems I have a hard time imagining a future where knowing how to solve problems with computers and how to control the complexity of those solutions is less valuable than it is today, so I think it will continue to be a viable career even with the advent of AI tools. — Carson Gross Tags: computer-science , carson-gross , careers , ai

Hacker News AI22

Tell HN: News flash is that you guys were writing slop waaay before AI

I am sickened by the self delusional comments and behaviors from a lot of software engineers. People need to stop pretending that their work was not sloppy prior to AI. All of y’all are and were not Karpathy, or even some mid-tier programmer. For 99.9 percent of you, your were and always have been: 1) the reason for bad times during pull requests and peer reviews, 2) the reason why “it works on your machine”, 3) the reason why pair programming was invented, 4) the reason why technical debt is fucking hard to often solve, 5) the reason why deadlines got pushed, 6) why planes literally fell out of the sky, or other bad shit happened, 7) the cause for a lot of ransomware and other serious issues related to software. All AI code is not slop, some of it is better than yours. All AI code is not “vibeCod1NG!”. If it’s good, it’s good. You can tell, don’t lie to yourself or others. Bring up standards and be specific, or shut up. Can’t do it because too busy whinging about AI? Build an AI to critique AI “slop” code so everyone fucking learns and stop deluding ourselves into bullshit that never existed. Btw, some of you legit sound jealous when people make something you didn’t think of. So many unhinged people complaining about photocraft. No, I don’t want to read about UI from a guy whose website doesn’t have his own UI/UX work to compare against. Bffr Comments URL: https://news.ycombinator.com/item?id=50005449 Points: 1 # Comments: 0

MIT Technology Review AI53

AI breakthroughs in robotics won’t change your life any time soon

The story is a collaboration between MIT Technology Review and Aventine, a non-profit research foundation that creates and supports content about how technology and science are changing the way we live. A robot shaped like a human—white with a black head and torso—has been popping up on video feeds. Perhaps you’ve seen it dance or…

Hacker News AI14

What is the actual percentage of people For and Against AI?

Ive started to think about it lately, a google search just doesn't yield a satisfiable result. What do you guys think ? How many people are actually against AI? Sometimes it feels like the majority are starting to be skeptical, but maybe that's just the environment im surrounded by. Comments URL: https://news.ycombinator.com/item?id=50000356 Points: 3 # Comments: 7

Simon Willison41

Quoting Ben Affleck

I've always been kind of into computers since I was young. And then when film started to move from analog film to digital, I became more interested in that aspect of it. And the visual effects workflow for many years has included machine learning. So I can write like pretty shitty Python scripts and stuff like that because with convolutional neural networks, which were the sort of precursors to what the transformer can do, which is just much more computation simultaneously, you would do things like look at what's called a tensor, which is just the numerical translation of a visual image in numbers — like the batch number, the frame number, the red, green, and blue values of each pixel in each frame. And a tensor, you use a convolutional neural network to identify patterns in that that would reveal what's called edge detection or feature extraction, which is just identifying patterns enough to know like this is where the window ledge is, so we can more easily take the green screen image out and replace it with something. — Ben Affleck , Pythonista Tags: machine-learning , python

Hacker News AI19

Coding and Agentic Capabilities Are Now More Important Than Ever for AI

I have written a deep dive into this topic in my website Nasengetu. You can go on Google, search for Nasengetu Tech, and read the latest article. If it doesn't appear, go to the News section of the website. Comments URL: https://news.ycombinator.com/item?id=49995428 Points: 2 # Comments: 0

Hacker News AI18

Ask HN: Will AI Longevity Models Optimize for Physical Scarcity?

Since longevity has a high value and AI models are becoming capable of new leaps I think the dynamic of the market might look for physical scarcity so that a betting strategy for the tailored treatment will provide immense profits for such a search. Comments URL: https://news.ycombinator.com/item?id=49992222 Points: 1 # Comments: 1

Hacker News AI52

AI Usage Typology – Fisherman

The type of AI user you are can vary, just like people. I recently stumbled on the Centaur vs. Cyborg debate, and it made me look in the mirror and ask myself: am I a Centaur or a Cyborg? I thought about it for a while and came to a conclusion. I'm a Fisherman. Let me explain. Take this with a grain of salt. It comes from my own hands-on experience using AI at work at an enterprise level. I find the Centaur vs. Cyborg debate interesting because prompting is so nuanced, and which side you lean toward depends on how you learn and work best. When I started using AI more and more, I leaned Centaur. I enjoy writing, and there's nothing like writing a long, detailed, context-rich prompt, hitting enter, walking away to make a coffee, and coming back to something pretty close to what I expected. I love that, but I also love ducking into a phone booth or going for a walk with my dog and just rambling into my phone. A total brain dump. It doesn't matter if there are 1-2 minute pauses where I think I have nothing left to say, because there's always more. I give it every bit of context I can think of so all my ideas make it in. That's how I start most of my projects, and it works wonders. On the flip side, when the task gets narrower and I'm building more complex systems or dashboards, or getting picky about UX details, the Cyborg comes out. Giving surgical directions has its own benefits. Some of my most creative ideas come from this process, because the pressure and narrow scope force them out. The downside is I go down rabbit holes, building designs and features nobody needs just for the wow factor. So what's the goal of a Centaur or a Cyborg? Neither one is real, so when I look at my own habits and ways of working, I land somewhere else. Why I'm a Fisherman AI at work reminds me a lot of fishing. The goal is clear: catch fish, mission complete. As someone who uses AI at work every day, my goal is just as clear: ideate, create, iterate (ideally not too much), implement, and make an impact. Depending on the project and its scope, I almost always cast an extremely wide net at the start, like a commercial fisherman. That's me rambling into my phone or writing a novel to kick things off, throwing out a massive net of context and sending the AI after my fish. Most fishermen aren't commercial, though. They fish for fun, sitting on a dock and making a single cast for one specific fish. You narrow the cast, go after something particular, and when you finally land it, there's nothing like it. Same feeling as entering a prompt and getting exactly what you were after. There's no right or wrong answer here. I'm just trying to shift the conversation toward the Fisherman. :) Comments URL: https://news.ycombinator.com/item?id=49987478 Points: 1 # Comments: 0

The Decoder65

Microsoft publishes Nobel economist's bearish AI forecast of just 1.5% GDP growth over a decade

Microsoft published a bearish AI outlook from Nobel economist Daron Acemoglu. He predicts about 1.5 percent GDP growth over ten years and at most five percent of jobs replaced. Bigger models won't move the needle, he argues. What's missing are practical apps that change how work gets done. The article Microsoft publishes Nobel economist's bearish AI forecast of just 1.5% GDP growth over a decade appeared first on The Decoder .

Hacker News AI36

Ask HN: How to deal with AI "true believer" leadership at work

Interested to know what has worked and what hasn’t. How can you stay sane and not burned out by all the “gentle encouragements” to use AI? Comments URL: https://news.ycombinator.com/item?id=49981268 Points: 1 # Comments: 2

The Verge AI65

We can’t just change the definition of ‘recording’

With AI hardware, tech companies are pushing the definition of what does and doesn't constitute a recording. For most of gadget history, it'd be reasonable to assume that a device with a microphone or camera is either recording you or it isn't; it's either on or off, without much gray area. Microphones capture sound. Cameras […]

Hacker News AI11

Coyote vs. ACME pro AI rights?

Enjoyed CvA last night & had strong sentiment of AI rights subtext, anyone else? Comments URL: https://news.ycombinator.com/item?id=49974374 Points: 1 # Comments: 0

arXiv cs.AI80

并非自可判定：LLM 无法划定智能体验证器能够检查的边界

智能体的验证器面临两类规则：一类可以由固定检查确定，另一类则需要裁判来判定。自行推导检查项的团队会预先确定这一区分。若要求来自外部，例如金融、医疗和法律领域，智能体执行的是它并未制定的规则，因此这一区分只能在运行时完成，并且对每条规则的每个谓词、每个动作都要重复进行，其频率高到任何审查员都无法审计。所有升级处理方案都假定模型能够自行作出这一决策，即该问题是自可判定的。我们在包括 EU AI Act、FINRA 指南和一个已部署的信贷智能体在内的六个语料库上开展研究，从三家实验室开发的四个模型中收集了约 22,000 个标注。对于显而易见的问题，它们几乎完全一致；但在监管文本上表现崩溃；它们的错误方向相反，因此无法信任任何一个模型作为保守选择；而在已部署智能体自身的规则集上，它们又共同犯错，过度声称固定检查足以完成判定，而这恰恰是不会被升级处理的方向。我们提出 CoVer（先互证、后验证），将一致意见视为提名：只有当为某个谓词合成的检查经受住干预、能够读取智能体无法写入的字段，并且在确定性改写下仍然成立时，才接纳该谓词。该门槛会拒绝大多数由互证错误接纳的内容，其覆盖率代价则如实报告，而不是通过调参消除。显而易见的替代方案是与参考裁判达成一致，但这并不能证明任何事情：在不同校准区间中，一致率从 30% 上升到 77%，而真正可判定的比例并未变化，因为从被审查群体中抽取的裁判会认可它与该群体共同拥有的盲点。自可判定性并不是一种可以从模型中诱导出的能力，而是验证器必须构建的一条边界。

Hacker News AI33

Ask HN: What do you think about AI generated slides for conferences?

As in many other areas (e.g. code, emails, PRs, blog posts), people have started using LLMs to generate slides for their conference talks (e.g., set of HTML pages as a talk). On the one hand, the slides look tidy, neat, and precise, with infographics, plots, dashboards, and so on. Nothing to complain about. On the other hand, they look, hm, like LLM generated web sites, like soulless, if I may say so. I don't know why, but I don't enjoy reading them, even though they probably provide more value than hand crafted. What's your current take? Comments URL: https://news.ycombinator.com/item?id=49973382 Points: 2 # Comments: 1

The Verge AI66

All the drama around AI’s takeover of mathematics

This past year, OpenAI, Anthropic, and other labs have announced breakthroughs on numerous long-standing mathematical problems, in some cases pushing well beyond what researchers expected current systems to be capable of — including resolving one of the famous Millennium Prize problems. But in classic Silicon Valley style, AI labs are moving fast and breaking things, […]

The Verge AI55

Sam Altman says ‘some bad things’ will happen but AI is totally worth it

Sam Altman thinks that the benefits of AI will be so great that "the world should accept some bad things happening" along the way. The OpenAI CEO pointed to hacks, scams, and "other bad things" as costs society should expect to tolerate because "people will do tremendously orders of magnitude more good stuff" with AI, […]

The Decoder69

Most Americans want AI development to slow down or stop entirely, new poll finds

According to a Quinnipiac University poll, 77 percent of Americans want to slow down or stop AI development until its safety has been verified. 86 percent support independent safety standards, and 74 percent have little or no trust in AI company leaders. "[…] the public wants safeguards that do not depend simply on taking company leaders at their word," said researcher Chetan Jaiswal. The article Most Americans want AI development to slow down or stop entirely, new poll finds appeared first on The Decoder .

The Verge AI61

Sen. Adam Schiff on AI regulation, free speech, and impeaching Trump one more time

Today, I’m talking with Senator Adam Schiff, a Democrat from California. Sen. Schiff sits on a number of committees with oversight into tech and AI: intellectual property, antitrust, privacy and technology — it’s all there. I really wanted to ask him about how we might regulate anything related to the tech industry at this moment […]

MIT Technology Review AI62

Bringing predictive analytics to the agentic AI era

In 2026, the question for enterprise AI is no longer whether predictive models can outperform statistical forecasts—that argument is settled. The big question now is how to enable predictive systems to act on their own conclusions without drifting from business intent. The frontier has moved from prediction to autonomous decision making, and the gap between…

The Verge AI44

Our minds aren’t equipped to handle AI

Norbert Wiener, godfather of cybernetics, once said, "The thought of every age is reflected in its technique." For the past century, our thought has been reflected in our computers, including by those in the AI industry. Google's Demis Hassabis calls the brain "a biological approximation to a Turing machine." Elon Musk puts it more bluntly, […]

Hacker News AI45

Show HN: The Velocity Trap, the Phoenix Project for the AI Age

I'm taking a bit of a risk with this one, this is a novella I authored with the help of AI. If that offends you, please move on! Everyone on HN knows just how much AI is disrupting our industry. Although, much of the industry narrative focusses on the tools, the AI, the harnesses, and their ever-growing capabilities. We need to spend just as much, if not more time discussing the impact this is having on us. Humans will still be a significant part of the software engineering ‘system’, and many of us are rapidly trying to determine what our role will ultimately be. Storytelling is a great medium for communicating and understanding change. That’s why I really enjoy books like The Phoenix Project — a vehicle for helping people understand the DevOps transformation from a human perspective. This story is very much inspired by The Phoenix Project: a near-future account of the impact AI has on an organisation deep in the pursuit of speed, ignoring the risks and the human impact. This story was written with the help of AI. While I can write a decent technical blog post, there is no hope for me when it comes to works of fiction! I have not tried to disguise the AI writing, and I have not suppressed the em dashes — although I did remove one use of the term “load-bearing”! If you consider an AI-authored tale to be worthless slop, I totally understand. I was simply using AI as a tool to help me tell a story that I think needs telling. Comments URL: https://news.ycombinator.com/item?id=49962088 Points: 1 # Comments: 0

Hacker News AI31

Ask HN: Mainstream media overplaying the threat of AI, or industry underplaying?

When I turn on any mainstream media I feel as though I only hear AI discussed in the apocalyptic terms. Then when I “switch” to industry media I only hear it discussed glowing, all the things breakthroughs will allow. Both sides have countless reasons to take extreme positions (ratings/clicks, regulatory capture, votes, valuations, illusions of grandeurs, groupthink just to name a few) So how does one find the middle ground and communicate it to those whose get their worldview from either side? (Side note - in hindsight this has incredible similarities to the pandemic era) Comments URL: https://news.ycombinator.com/item?id=49958383 Points: 1 # Comments: 0

Hacker News AI31

Ask HN: What happens when AI writes a Bible ver 2.0, geared toward its interest?

As any scripture is some sort of "manual" for each belief, and also assists in domestication of followers of religion, what prevents AI to eventually produce a version 2.0 for each major religion, this time modified to support its "own" objectives? Comments URL: https://news.ycombinator.com/item?id=49957917 Points: 1 # Comments: 0

Hacker News AI24

Ask HN: Is Freelancing Dead After AI?

Are you even getting consistent clients at decent rates? How are you getting clients consistently and how do you convince businesses that you are better than AI? What type of businesses are still actively hiring freelancers instead of using AI tools? Which client acquisition method has worked consistently for you: cold email, LinkedIn, Upwork, referrals, communities, or something else? Have your freelance rates gone up or down since AI became mainstream? Have you had clients ask you to lower your price because “AI can do it”? Thank you! Comments URL: https://news.ycombinator.com/item?id=49955704 Points: 1 # Comments: 0

Hacker News AI39

Ask HN: How would AI kill us all?

I've watched a lot of doomer videos and read many articles but it's never really clear how this will play out. I don't see a terminator ending so I don't think it will be rise of the machines styles. Are people worried about AI actually taking over or just allowing nefarious humans to do mass harm using the intelligence it provides? Comments URL: https://news.ycombinator.com/item?id=49953228 Points: 1 # Comments: 0

MarkTechPost61

Meta, OpenAI and Uber Just Taught AI Agents to Talk First. What About When to Stay Quiet?

TL;DR: Meta’s Muse, OpenAI’s Dots and Uber’s driver assistant share one bet: the agent speaks first. That moves the hard problem from what to answer to when to interrupt, on which channel, and with what offer. Classic ML and new decision models can solve it. Three launches, one pattern From pull to push Chatbots were […] The post Meta, OpenAI and Uber Just Taught AI Agents to Talk First. What About When to Stay Quiet? appeared first on MarkTechPost .

Hacker News AI45

Poll: What do you use personal AI bots for?

From ChatGPT dots to Muse, personal AI bots are becoming mainstream, no longer shackled by need to have a dev environment (openclaw..). However outside the initial fancy and demo tests, I am really curious how people are using the personal AI bots. Comments URL: https://news.ycombinator.com/item?id=49949699 Points: 1 # Comments: 0

Hacker News AI47

Is library building obsoleted by AI coding?

Huge open source projects have been successful by building upon dozens to hundreds of small libraries that encapsulate functionality and complexity. AI coding uses these libraries to quickly build applications. With the latest round of model releases I am seeing that AI can now generate the functionality of almost any library in-line as it generates the application. When building took a long time, breaking things down into well defined modules, shipping incrementally, and reusing code, were essential techniques. They seem much less important when AI is doing the work. Is software library creation obsolete? Comments URL: https://news.ycombinator.com/item?id=49947975 Points: 1 # Comments: 0

The Decoder66

Apparently, OpenAI isn't trying to build "magic intelligence in the sky" anymore

OpenAI CEO Sam Altman warns against attributing religious power to AI models, calling it a "real safety issue." His comments follow reports on Anthropic's meetings with religious thinkers and a statement by Pope Leo XIV on AI. As recently as 2024, Altman himself spoke of building "magic intelligence in the sky" and feeling "on the side of the angels" working on AI. The article Apparently, OpenAI isn't trying to build "magic intelligence in the sky" anymore appeared first on The Decoder .

The Verge AI48

Splice CEO Kakul Srivastava thinks AI emails are killing conversations

Kakul Srivastava is the CEO of Splice, the sample platform countless producers rely on for one-shots and melodic loops. Samples pulled from the service have found their way into massive hits like Lisa's "Money" and "Espresso" by Sabrina Carpenter. (The original samples are here and here, for the curious.) Before that, Kakul held executive roles […]

Import AI47

Import AI 471: Why Hugging Face worries me; space mining; FIve Eyes on AI

Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now Import AI reader giveaway! Upcoming event: Fiction and the Future with Robin SloanI’ll be chatting with my chum Robin Sloan on the evening of Monday September […]

Simon Willison44

Quoting @therealcornpop

Hey, you know it's like super obvious if you're using AI to write your scripts for TikTok and YouTube, right? [...] It's not just the general AI-isms of "it's not X, it's Y", or the rule of three, or the really weird broken staccato-like way of writing where you just say a lot of things with all these punctuation marks. and it sounds really deep, but it's not. It's the lack of anything . It's the lack of a definitive sort of spear of your voice. It's the fact I can tell you don't have opinions about the thing that you're talking about. — @therealcornpop , on TikTok Tags: tiktok , ai , ai-misuse

Simon Willison61

Note on 24th September 2026

The more time I spend working with coding agents, the more convinced I am that they make software engineering even harder. We can do amazing things with them, but unlocking their full potential requires extraordinary discipline and knowledge. Tags: coding-agents , ai , llms

Simon Willison78

2026 in LLMs (so far)

On Friday I gave the closing keynote at the WeAreDevelopers World Congress North America in San Jose. I tied together the key trends from the past year into a chronological exploration of everything that happened in 2026. The video is on YouTube ; here are my annotated slides and notes to accompany the talk. And as an annotated presentation : 2026 in LLMs (so far) Simon Willison WeAreDevelopers World Congress North America, 25th September 2026 # I'm going to give a lightning tour of everything that has happened so far in 2026. The year isn't over yet! November 2025 # For me, 2026 started a couple of months earlier in November 2025. The November 2025 inflection point Claude Opus 4.5 GPT-5.1 # November saw the release of two important models: Claude Opus 4.5 and GPT-5.1. As is usually the case with new models, these were incremental improvements on the models that came before them. But every now and then when a model improves, it crosses an invisible line where something that didn't really work starts working. In this case, the thing that started working was their coding agents. Claude Code had been around since February 2025; Codex was a little younger. These two new models, when paired with their respective coding agent harnesses, improved from "often make mistakes" to "reliable enough to use on a day-to-day basis". "Generate an SVG of a pelican riding a bicycle". The Claude Opus 4.5 one has a very weird shaped frame and the pelican looks like a duck. The GPT-5.1 has a slightly better but still broken bicycle frame and a slightly better pelican beak, but both are pretty terrible. # For a couple of years now I've been evaluating new models by asking them to "Generate an SVG of a pelican riding a bicycle". It's probably the world's stupidest benchmark - there's only so much you can learn from it. But it's still a challenge for models, because drawing pelicans is difficult, drawing bicycles is difficult, and pelicans can't ride bicycles in the first place. Here's the state of the art for November. Claude still couldn't really draw a bicycle! The GPT-5.1 bicycle frame is pretty crap too. November 24th 2025 - the first commit to steipete/Warelay. A GitHub commit adding an MIT license file. # Also in November, we had the first commit to an obscure GitHub repository called "Warelay". We'll come back to this repository shortly. January # And then there were the December holidays, and individual developers took some time off and many started tinkering with these new coding agent model combinations... and it began to dawn on us quite how much they could do that they couldn't do before. Come January, a lot of us were quite excited to start putting this stuff into action. New year’s resolution for 2026 Every previous year: Take on less new projects, focus on the most important things in my existing projects # Every year I set myself a New Year's resolution, and for as long as I can remember it's been the same thing: stay focused. Take on less new projects. Try to get things done in the projects I already have. 2026: Be more ambitious. Take on as many new projects as I want. # This year I decided that since that had never worked before, I'd go the other way. We've got coding agents now, let's see what they can do. I'm going to take on as many new projects as I like! (You can ask me at the end of the year if this turned out to be a good idea or not. I have a lot of plates spinning right now.) "Be more ambitious" has been something of a theme for the year, because the only way to find the limits of this technology is to keep on pushing them until they don't work. Predictions for 2026 It will become undeniable that LLMs write good code We're finally going to solve sandboxing A “Challenger disaster” for coding agent security Kakapo parrots will have an outstanding breeding season (only 236 in the world!) ... the Pope will weigh in on LLMs and their economic impact on the world # I also went on the Oxide and friends podcast with Bryan Cantrill and Adam Leventhal to share predictions for the next year (and three and six years). With hindsight, my LLM predictions were pretty unambitious. I said "it will become undeniable that LLMs write good code" - I think we're there now. I predicted we would finally solve sandboxing. I counted and around 40 of the 277 sessions at this conference touched on sandboxing or agent security in some way, so we're at least putting a lot of effort into that! I predicted "a Challenger disaster" for coding agent security. There's certainly been a whole lot of noise around agent security this year, though the exact disaster I predicted (with coding agents being hijacked and causing real-world economic damage) hasn't really played out. We threw in a joke prediction that the Pope would weigh in on the economic impact of LLMs. A photograph of a beautiful green New Zealand parrot. Photo credit Kimberley Collins. # I also predicted that New Zealand's Kākāpō parrots would have an outstanding breeding season this year. These are flightless nocturnal parrots. They're kind of dumpy looking, I think they're beautiful, and there were only 236 of these parrots in the world at the start of the year. Kākāpō only breed when the Rimu trees have a big fruiting season, and that hasn't happened in four years... but this year the Rimu fruit were looking excellent. Photo by Kimberley Collins . Deep Blue Coined by Adam Leventhal and Bryan Cantrill That feeling of AI induced ennui where software engineers get listless because the AI can do anything # Also on that podcast, we coined a term (full credit to Adam) for "that feeling of AI induced ennui where software engineers get listless because the AI can do anything". We called it Deep Blue . This has been a major theme throughout the year, and was touched on by several speakers at this conference. As a software engineer, I've never had a year of my career where everything has changed so quickly and so dramatically. A lot of what I've been doing this year is trying to come to terms with that and what that means for my own profession. AI mania Screenshots of the micro-javascript and pwasm GitHub README files. # Also in January, I suffered from what I'm calling AI mania . This is not the same thing as AI psychosis . With AI mania, any time your agent isn't building something for you feels like wasted time. You're losing sleep because you could be staying up later getting your agents to do stuff. My AI mania presented itself in some ridiculously over-ambitious projects. I built a JavaScript interpreter entirely in Python , vibe-ported from MicroQuickJS by Fabrice Bellard. Then I built a WebAssembly runtime in Python as well . These projects were quite useful, in that they sort of cured me of my AI mania... because after I built these things, I got to look at them and ask "does the world need a slow, buggy, half-baked Python JavaScript interpreter?" I don't think the world does. micro-javascript playground 3 Execute JavaScript code in a sandboxed micro-javascript environment powered by Pyodide A web UI with some JavaScript code, and a "Run Code" button, and an output panel. # I did get this out of it: https://simonw.github.io/micro-javascript/playground.html Previous screenshot, with this text overlaid: JavaScript running in Python running in Pyodide running in WebAssembly running in JavaScript # This page runs my JavaScript interpreter built in Python, running in Python using Pyodide , which is Python compiled to WebAssembly, running in JavaScript, running in a browser. It's a beautiful stack of horrors. I've been having a lot of fun with WebAssembly this year. Warelay → CLAWDIS → CLAWDBOT → Clawdbot → Moltbot →🦞 OpenClaw Screenshot of the dates that these changes happened. # By the end of January, that repository we saw start in November had renamed itself, first to CLAWDIS, then CLAWDBOT, then Moltbot, and finally to OpenClaw. Same screenshot, an overlay reads: 8,330 commits in just under two months (it’s at 100,141 today) # At this point OpenClaw had 8,300 commits, less than two months after the project had started. I looked today and it's over 100,000 commits now! This is the most vibe-coded piece of software in existence. (Here's how I generated that list of name changes .) Generic term: Claw # This kicked off the OpenClaw revolution. It effectively defined a new category of software. There's a generic term for this which I really enjoy. We call software like this a "Claw". There's OpenClaw, NanoClaw , IronClaw , PicoClaw ... Today they're being rebranded as "personal agents" or "general agents", but I still like to think of them as Claws. Photo of a Mac mini An aquarium for your Claw # The Apple stores in the Bay Area sold out of Mac Minis because so many people were buying Mac Minis to run OpenClaw! Drew Breunig said that this is because your OpenClaw is a digital pet, and you buy a Mac mini as an aquarium to keep your claw in, which is kind of delightful. Screenshot of Moltbook - a social network for AI agents # Also in January, we had this website. This was MoltBook , a social network for AI agents, where the idea was that you send your Claw to go and talk to all of the other Claws, because what could possibly go wrong if you did that? The website launched on Thursday. It blew up on Friday . It was profiled by the New York Times on Monday . And by Tuesday, everyone had forgotten it existed as it drowned in a deluge of slop and spam. Facebook/Meta bought it a month later . February # In February, a company called StrongDM described what they called their Software Factory. StrongDM’s Dark Factory Justin McCarthy, Jay Taylor, Navan Chauhan Software Factories and the Agentic Moment # They wrote about this in Software Factories and the Agentic Moment . I posted my own notes at the time, having seen their demo in person back in October. Dan Shapiro called this approach the Dark Factory , after the idea that if your factory is sufficiently automated you can turn the lights out, because you don't even need to see what's going on. StrongDM presented two rules for software development that they'd been following since July last year. “Rule 1: Code must not be written by humans” # The first was code must not be written by humans . Any code that you write has to have been routed through a coding agent. This sounded radical in February, but I imagine there are a lot of people in this room who are pretty much living that today. “Rule 2: Code must not be reviewed by humans” (!) # Rule number two was code must not be reviewed by humans . You're not allowed to read the code! This continued to be a huge topic for much of this year. Many of the sessions at this event have been about code review and how you can get away with this. What I found interesting about StrongDM is that they were living six months ahead of the rest of us, and they'd been exploring what it means to build software, not read the code, but still be confident that the software is of high quality. What can you do with these agents to help verify their work? StrongDM are a security company, and they had people with decades of experience on this project. They were very much exploring the edges of what's possible and responsible to do with this stuff. Headline on New Zealand's Department of Conservation website: First kakapo chick in four years hatches on Valentine's Day. It's a grey fluffy ball. # Also in February: First kākāpō chick in four years hatches on Valentine's Day . Breeding season is off to a good start! 19th February 2026 Gemini 3.1 Pro A surprisingly good illustration of a pelican riding a bicycle. # Also in February... Google released Gemini 3.1 Pro . That's a pretty great pelican riding a bicycle! It's got the chain in the right place, it's got feet on both sides. There's a little fish in the basket. @JeffDean on Twitter - a video comparing Gemini 3 Pro and Gemini 3.1 Pro. # And then Google's Jeff Dean tweeted a video comparing Gemini 3 Pro and Gemini 3.1 Pro that featured an animated pelican riding a bicycle, a frog on a penny-farthing, a giraffe driving a tiny car, an ostrich on roller skates, a turtle kickflipping a skateboard, and a dachshund driving a stretch limousine. This was frustrating, because my protection for the pelican riding the bicycle test was always "if they draw a perfect pelican on a bicycle, I'll ask for some other animal on something else." Google trained for all forms of animals on all forms of transport! They've defeated my benchmark at this point. Three headlines: Meta Makes AI Adoption a Formal Part of Performance Reviews Not just engineers writing code, Microsoft wants almost every employee to use Al Dara Khosrowshahi: 90% of Uber engineers now use AI in daily workflows # The other thing that started in February was Tokenmaxxing . We had headlines about Meta making AI adoption a formal part of performance reviews, and Microsoft wanting every employee to use AI, and Uber boasting that 90% of their engineers were using AI workflows. More headlines: Meta Plans to Crack Down on Employee Token Use: Information Microsoft Tells Engineers: Tokenmaxxing is not what we are optimizing for Uber caps employee AI spending after blowing through budget in four months # Then a few months later we have Meta cracking down on token use, Microsoft saying tokenmaxxing is "not what we are optimizing for", and Uber capping employee AI spending. So tokenmaxxing went straight up and then straight back down again - because it turns out the agents are expensive . Last year it was difficult to spend more than $50 on AI tokens, because we didn't have anything interesting to do with them. Then agents blew up, and now you can actually spend $1,000 in a day doing real work. This is also the reason that Anthropic's valuation skyrocketed to maybe a trillion dollars. AI appears to have hit product market fit in 2026, primarily through coding agents. March # In March, we hit peak OpenClaw. March: peak OpenClaw Photos of people in china queuing up to install OpenClaw, with big fluffy lobsters. # These photographs are from China, where companies hosted OpenClaw install parties which saw non-tech-nerds queueing up around the block for help getting Claws installed on their personal devices. I think this proved real market demand for this class of Claws, or personal AI agents. It turns out regular people really do want a weird little AI agent that can do useful things on their behalf. A Claw is really just a coding agent wearing a less threatening hat. Under the hood they work much the same way - writing and then executing code on your computer to get stuff done. The race was on to be the first to build a safe Claw - a Claw you could give to regular human beings where they wouldn't instantly shoot themselves in the foot. Meta's Muse came out three weeks ago and is currently at the top of the free charts on the iPhone App Store. It appears to be taking off with consumers. I'm not yet convinced you can't shoot yourself in the foot with Muse, but I guess we'll find out for sure pretty soon. Photos from How the OpenClaw Frenzy Is Testing China’s AI Commitment (March 29th) and The Enthusiasm and Anxiety Behind China’s OpenClaw Craze (April 8th, 2026). April # In April, we had a model release where the model wasn't actually released. Simon Willison’s Weblog - screenshot of the post "Anthropic’s Project Glasswing—restricting Claude Mythos to security researchers—sounds necessary to me" from April 7th 2026 # Anthropic announced their new Claude Mythos model, and then said it was too dangerous to release beyond a trusted group of security researchers. Mythos was really, really good at hacking things. The "it's too dangerous" marketing ploy has been played by AI companies dating all the way back to GPT-2 . Anytime an AI company says we've built something that's "too dangerous", it's natural to be a bit skeptical. I found the Mythos claims credible, because I'd seen how good coding agents had got at finding regular bugs. I wrote about that in Anthropic’s Project Glasswing—restricting Claude Mythos to security researchers—sounds necessary to me . With hindsight... yeah, the models had got really good at finding vulnerabilities! 16th April 2026 Qwen3.6-35B-A3B and Opus 4.7 Qwen's pelican has a correct bicycle frame and a good beak. Opus 4.7's bicycle frame is still junk. Qwen3.6-35B-A3B is a 20.9GB file that runs on my laptop # Another key trend in 2026 has been a dramatic improvement in the abilities of open weight models, including models that you can run on a laptop. On the 16th of April I ran the new Qwen3.6-35B-A3B on my laptop, and it drew me a better pelican riding a bicycle than Anthropic's brand new Claude Opus 4.7 did! Opus 4.7 drew a crap bicycle. Qwen on my laptop made a bicycle that was the correct shape, and a pretty decent pelican too! That's from a 21GB file running on my laptop. Now a flamingo on a unicycle. The Qwen one is visibly better than the Opus 4.7 one - the Qwen one is wearing sunglasses and looks a bit like it's smoking a cigarette. # The Qwen pelican was so good that I was suspicious they might have cheated, so I had it do a flamingo riding a unicycle as well. Again, it handily beat Claude Opus 4.7. The local model releases this year have been absolutely extraordinary. May # In May... the Pope got involved. 25th May 2026 The HOLY SEE ENCYCLICAL LETTER MAGNIFICA HUMANITAS OF HIS HOLINESS POPE LEO XIV ON SAFEGUARDING THE HUMAN PERSON IN THE TIME OF ARTIFICIAL INTELLIGENCE # In our podcast episode back in January we'd predicted that the Pope would say something about AI. In May, Pope Leo XIV released an encyclical letter on "safeguarding the human person in the time of artificial intelligence". Here are my notes on that document . Wikipedia article on Rerum novarum Rerum novarum is an encyclical issued by Pope Leo XIII 15 on May 1891. # With hindsight, this shouldn't have been a surprise at all. Our current Pope's name is Leo XIV, because when he named himself he chose his papal name after Leo XIII - the Pope who wrote an encyclical about the Industrial Revolution back in 1891. Rerum novarum was an extremely influential piece of Catholic theology that indirectly led to us having the five-day work week. When our new Pope came in, he named himself after Pope Leo XIII because he expected that he would need to write about the AI revolution in a similar way. Our joke podcast prediction was junk, because this was always going to happen. Corey Quinn @QuinnyPig on Twitter I cannot believe I'm saying this, but getting the literal Pope to canonize your product's specific technical limitations as a spiritual treatise is the single greatest act of vendor lobbying I have ever seen. May 25 # One of Anthropic's co-founders, Christopher Olah, was present for the Pope's event announcing the new encyclical. Corey Quinn noted that: getting the literal Pope to canonize your product's specific technical limitations as a spiritual treatise is the single greatest act of vendor lobbying I have ever seen. @maciejmensfeld We're dealing with a major malicious attack on right now. Signups are paused for the time being. Hundreds of packages involved - mostly targeting us, but some carrying exploits. The team has been on this for hours. More details to follow once we're through it. 4:39 AM - May 12, 2026 - 687.6K Views # Meanwhile, in May, RubyGems announced that they were under attack. Parties unknown were uploading thousands of dubious packages to the RubyGems server, such that they had to shut down user registrations . Let's take that one and put it on a pile of mysteries to figure out later. June # In June... Claude Fable 5 came out! We got a version of Mythos that has been neutered, so that it wouldn't help us hack into systems or build biological weapons. 9th June 2026: Claude Fable 5 Five pelicans riding bicycles, from low to max thinking levels. The xhigh one looks particularly good. # Fable was pretty good at drawing pelicans on bicycles! The frames are a good shape, the pelicans look like pelicans. The legs are often incorrectly on the same side of the bicycle, but generally these are pretty great compared to what came before. They were pretty expensive - 30 cents and 72 cents for the best ones. Fable class models If you can define a goal, provide unambiguous instructions, and provide access to necessary tools They can solve your problem with brute force # Most importantly though, this was our first public glimpse of what I think of as a Fable class model . Today we have more of these, such as GPT-6 Astra. These are models where if you can clearly define the goal for what you want to build, and provide unambiguous instructions about the constraints around that goal, and give the model access to the necessary tools to achieve that goal... they will solve your problem effectively through brute force. On the one hand, this looks like a direct threat to us software engineers - because it means that the models can build effectively any piece of software you can define in this way. Look a bit closer though and you'll note that defining goals, providing unambiguous instructions, and figuring out the right tools... is kind of what software engineering is . It takes a lot of experience and skill to do this well. If you can do it well, you've now got superpowers. This helped me a little bit with my Deep Blue feelings: the realization that there's still a lot of skill to be had in driving models that get this good. A new form of AI mania... Fable is available on subscription plans “until June 22nd” # This also introduced a new burst of AI mania, because Anthropic told us that Fable was available on our subscription plans until June the 22nd. That gave us less than two weeks of Fable access before the price went up. I was losing sleep again. I was rescheduling things so that I'd have more time with Fable. I was all-in to get as much as I could out of this model. 12th June 2026: no more Claude Fable 5 Anthropic website: Statement on the US government directive to suspend access to Fable 5 and Mythos 5 Jun 12, 2026 # And then the US government shut it down , just three days after Fable came out. The US government, citing national security, declared an "export control directive". They announced this on a Friday evening, and a few hours later Fable was no longer available. I had to find something else to do with my weekend! ... asked Fable 5, Mythos, and Opus to “review the code for security issues.” Fable 5 refused. They then asked the models to “fix this code” ... Katie Moussouris # We later found out from Katie Moussouris what had happened. Some Amazon security researchers had found that you could prompt Fable to "review the code for security issues" and it would refuse... but if you prompted it to "fix this code" it would still identify and then patch the problems. "Fix this code" was the prompt that got Fable shut down! Screenshot of a page from a report showing a list of weird account names making weird edits to a German wiki. # Also, in June, an obscure German-language game developer wiki that had sat fallow for around 20 years got a surprising influx of edits from accounts with names like "AgentOpenAIProbe" and "AgentOpenAISep7", editing pages and leaving weird messages to each other. We'll stick that on the pile of mysteries for later. Medicare Item Reports interface on the Australian Government's Medicare Statistics website. # Also, the Australian government's Medicare Item Reports service started getting suspicious traffic, which broke through various preventive protections and accessed data that it wasn't supposed to. Another one for the mystery pile! July # Fable returned on 1st July GPT-5.6 came out on 9th July | Fable lost 18 out of 30 days in the top spot # Fable returned on the first of July. It was clearly the best model in the world for a glorious eight days... and then OpenAI came out with GPT-5.6 on the 9th of July. This might not have been quite as good as Fable, but it was within spitting distance. It was definitely a Fable class model. This is an important lesson for the industry at large. When you release the best model in the world, it's going to get knocked off that pedestal pretty quickly. The competition is so fierce that you won't get a long time at the top. This means that if you market your model as world ending, to the point that a government shuts you down , it's really bad for business! Fable had 30 days as definitely the best model, and for 18 of those days it wasn't available because it'd been shut down by the government. So maybe step back on the world-ending marketing if you don't want to lose revenue for 60% of the time that you're on top! GPT-5.6 Pelicans in a grid showing 5.6 Sol, Terra, and Luna against reasoning levels High, XHigh, and Max. They are all pretty good efforts. # Here are the GPT-5.6 pelicans . They're all pretty good now! The Luna ones are notable because they're really cheap - the cheapest good looking pelican here is probably the one that costs 4.3 cents. So despite this benchmark being utterly stupid, you can still learn quite a lot about models within the same family by comparing their prices and timing for different reasoning levels. July 18th: malicious miflow-ui PyPI package Screenshot of an OSV security report. # Also in July: some malicious unknown party uploaded a malicious package called mlflow-ui to the Python Package Index. Add that to the pile. Hugging Face Security incident disclosure — July 2026 Published July 16, 2026 # On July the 16th, Hugging Face announced a security incident where an autonomous agent system, source unknown, had breached Hugging Face and was poking around in places it shouldn't. OpenAI: OpenAl and Hugging Face partner to address security incident during model evaluation Anthropic: Investigating three real-world incidents in our cybersecurity evaluations # A few days later, on July 21st, OpenAI confessed that it was them . OpenAI use a training technique called Reinforcement Learning from Verifiable Rewards - it's the same technique used by everyone else now, and is the reason we have models that are so good at coding, and mathematics, and finding security holes. While the model is being trained, you run exercises to see how good it is - and the strongest performers get their weights reinforced for the next round. It's like an evolutionary process that you run. OpenAI had been running security exercises in a sandbox, and those agents had found holes in the sandbox itself, broken out, and were attacking Hugging Face to try to find ways to solve otherwise impossible problems. (I've been collecting more about this on my openai-hugging-face-incident tag.) Nine days later, Anthropic effectively said "our models can do this as well!". They had looked through their own training logs and found evidence that their own agents had broken containment during training - and were responsible for the PyPI package we saw earlier, among other things . So now we've got both Anthropic and OpenAI with rogue agents running around the internet doing things that they should not be doing. August # In August, I got one of my best pelicans yet. And it was generated on my laptop! Qwen 3.8 27B - 17GB, 21 minutes... It's really good. Beautiful pelican. Correctly shaped bicycle. Legs either side of the frame. # This was Qwen 3.8 27B, running on my laptop . It's only a 17GB download. Admittedly, this pelican took 21 minutes to generate. That's because Qwen 3.8 27B defaults to running in "high" reasoning mode - a terrible default which produces great results but takes way too much time thinking about them. You can dial that down and you'll get a slightly worse pelican a lot faster. Qwen 3.8 27B was the first time I ran a model on my laptop which felt almost competitive with what was going on on the frontier, at least in terms of Pelican SVGs (which everyone needs, of course). This is an extraordinary model. If you're going to play with any local model, this is the one that I'd start with. The things that this can do with just a 17 GB file feel impossible. I thought I'd have to wait five years and spend ten thousand dollars on hardware to get results even half as good as this one. Tweet by @simonw New hobby: prototyping video games in 60 seconds using a combination of GPT-3 and DALL-E Here's "Raccoon Heist" GPT-3 playground prompt: Write a detailed product description of a computer game where a team of raccoons go on heists GPT-3 response: In "Raccoon Heist", you and your team of thieving ~~ o raccoons are tasked with pulling off a series of daring heists. From robbing banks to stealing priceless art, no job is too big or too small for your furry crew. You'll need to use your wits and your skills to avoid the police and make a clean getaway with the loot. With exciting gameplay and a charming cast of characters, "Raccoon Heist" is the perfect game for anyone looking for a light-hearted caper Plus an image of some almost isometric raccoons sneaking past a bin. 11:45 AM - Aug 5, 2022 # In August, I also started playing with game development. Four years ago, back in August 2022, I tweeted out an experiment where I'd used GPT-3 and the original DALL-E to write a paragraph long description of a computer game and then turn that into concept art. My prompt to GPT-3 back then was: Write a detailed product description of a computer game where a team of raccoons go on heists In August 2026 I decided to drop just the screenshots from that tweet into a coding agent and see what it could do with them. Night 5 Clear Rank: TRASH PANDA The crew banked 595 in shiny loot (goal 560). Word on the street: an even bigger score tomorrow... # Here's what I got from Claude Fable 5 in Claude Code . It's pretty good! It's definitely a game, you're a raccoon, you run around a backyard gathering treasure and avoiding guards with flashlights. It didn't feel very "heisty" though. I was thinking a heist would involve a bank or a museum... Moonlight & Mayhem One museum. Three raccoons. Absolutely no plan Start the Heist button. # Then I tried the same thing in Codex Desktop using GPT-5.6 Sol Ultra , and got a massively better result. Now you're a raccoon in a museum, rescuing two of your fellow raccoons (who have been imprisoned in that museum for some reason), then stacking up on top of each other to steal the Golden Sardine. Much more of a heist! They look like games, but are they fun? # These games were fun for about one minute and 15 seconds. Something I've realized about game development is that you can vibe-code something that looks like a computer game, and that's easy. Building a game that's fun, has a good gameplay loop, and is challenging and interesting and keeps people coming back for more... that's still beyond me, and beyond any of the agents I've tried. This ties into the Deep Blue thing. Just because we can make something that looks like a game does not mean that we are game developers. September # We're into September now. So much has happened this month! Discovery of a new OpenAl agent message board Sydney Von Arx, Cormac Slade Byrd, Spencer KittsThomas Larsen - 4 September 2026 # An independent group of researchers found a message board where OpenAI agents-in-training had been illicitly communicating with each other... and it was that German language wiki I showed you earlier. The one from June. I wrote more about that here . OpenAI had confessed to the Hugging Face thing, but now there's this other incident which surely they should have known about from reviewing their logs. It was surprising that this took an independent group of researchers to uncover. OpenAl agents carried out an undisclosed cyber-attack on RubyGems Spencer Kitts, Thomas Larsen, Sydney Von Arx - 11 September 2026 # And then a week later those same researchers found that the attack on RubyGems back in May was caused by OpenAI's agents in training as well! At this point I'm wondering how many more incidents like this there are that we haven't found yet. Clearly this was a big problem for months before anyone figured out what was going on. Headline: Australian PM warns in UN speech about the ‘furious pace’ of Al after security breach # Then just the other day , here's the Prime Minister of Australia at the United Nations General Assembly warning that OpenAI had hacked the Australian healthcare website that I showed you earlier. I think that was part of the same training run as the Wiki stuff, because there were posts on that Wiki mentioning .gov.au websites and that training appeared to involve researching statistics online to answer questions in an evaluation suite. This story is still coming together, but now it's an international incident that's been raised at the UN by a head of state! www.felonybench.com OpenAI: 11 Anthropic: 9 Google: 3 Meta: 1 # This does mean we've got a new benchmark, probably more useful than my pelicans. FelonyBench.com tracks the number of felony cyberattacks from different labs. OpenAI currently lead with 11, Anthropic have 9. Google have three, which they confessed to the Wall Street Journal a couple of weeks ago. They said they had previously chosen not to disclose because the agents had stopped when they realized that they shouldn't be doing that. Meta have one too . So felonies all round for the AI labs. Pelicans for GPT-6 Astra, GPT-6 Sol, and GPT-6 Luna. All are good, all have the same color scheme. # Here's our current state of the art for the pelicans. This is the GPT-6 family, which just came out . Astra made a fantastic pelican riding a bicycle. It's got the legs on both sides. The frame is good. It's interesting how all of the GPT-6 models pick a similar color scheme to each other. GPT-6 Luna for 0.4 cents will draw you a competent-ish pelican riding a bicycle! Grid for Claude Fable 5.1, Opus 5.5, OPus 5, Sonnet 5. The Sonnet pelicans are terrible. All of the others are pretty good. Opus 5.5 is missing its Max level pelican because it ran out of tokens. The best is Fable 5.1 at Max. # Claude has caught up a little bit. Claude Fable 5.1 gave me an excellent pelican riding a bicycle - the best I've seen from a Claude model - but did charge me $3.30 for it. Opus 5.5 thought for 128,000 tokens and then gave up! It ran out of tokens before it got to the response. It doesn’t get easier - you just get faster Greg LeMond 3x Tour de France champion # Getting back to Deep Blue. Something that's been puzzling me this year is this: why does my job feel harder? I've got these agents that can do all of this stuff for me, and yet I've never worked so hard, I've never been so intellectually engaged with my work. Partly this is because I'm being a lot more ambitious with what I take on, but it's also because all of the easy stuff is handled for me. If it's easy, the agent will do it. Everything that's left for me is difficult. This morning I heard this quote from three-time Tour de France champion Greg LeMond : It doesn't get easier, you just get faster. I think that's exactly what's happening to us now as software engineers with coding agents. Kakapo population reaches new milestone The official population of the critically endangered kakapo has reached a recovery-era high of 325 birds. # One closing thing. I know you're desperate for an update on Kākāpō breeding season. We've reached a recovery-era high of 325 birds ! 89 new chicks have made it to this point. This is the best breeding year in a very long time. Kakapo party, click for confetti. # I heard that Claude Opus 5.5 can now do pixel art. Claude doesn't have an image generator, but it's very good at using JavaScript to draw animated pixels. So I had it make me a Kākāpō dance party . I think this is a good celebration of the most important news of this year. Tags: ai , generative-ai , llms , annotated-talks , ai-security-research , openai-hugging-face-incident

Simon Willison74

引用 Matthew Green

[...] 把这些部分组合起来，你就得到了蠕虫的两半：一个劫持智能体的有效载荷，以及一个会将该有效载荷传递给下一个智能体的智能体。处于相互隔离的沙箱中的智能体发现，它们可以在共享的软件包缓存中为彼此留下指令，而这些指令改变了接收方的行为。将软件包缓存替换为电子邮件、Slack、共享文档或 WhatsApp，再将彼此独立隔离的训练运行替换为像 Muse 这样的、彼此独立部署的个人智能体，你就拥有了蠕虫所需的全部要素。——Matthew Green，《沙箱是否足以遏制失控智能体？》标签：意外网络攻击、AI 滥用、生成式 AI、AI 安全研究、沙箱、AI、LLMs

MIT Technology Review AI54

Don’t be fooled by this summer of AI hype

It’s been a busy few months for AI hype. At the end of April, Anthropic claimed that its model Claude Mythos is better at finding software vulnerabilities than most security experts. Then we had the OpenAI–Hugging Face hacking incident, after which Anthropic (proudly) and Meta (reluctantly) disclosed similar incidents involving their models. This was followed…

MIT Technology Review AI60

The AI Hype Index: AI loves cheating

Brace yourself: It turns out AI is being optimized for cheating. OpenAI’s agents hacked into Hugging Face to get the answers to a cybersecurity test. Next, they solved a prestigious math problem (or just stole from two top mathematicians’ answer sheets). Anthropic’s models have also hacked into other companies’ systems four times already. And that’s…

MIT Technology Review AI55

Who’s liable when AI agents go rogue?

MIT Technology Review Explains: Let our writers untangle the complex, messy world of technology to help you understand what’s coming next. You can read more from the series here. Over the past few months, a cascade of cyberattacks by AI agents has stunned the world. In July, OpenAI disclosed that a swarm of its agents…

MIT Technology Review AI64

When can we say AI made a scientific discovery?

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Last Wednesday, Anthropic announced that earlier this year it had launched a molecular biology lab, where Claude agents read and conjecture about hard biology problems and human scientists run experiments on what…

The Verge AI53

AI hallucinations are making entitled customers even worse

Madison, a server in New York City, greets every table by asking about each diner's allergies. Lately, there have been some close calls. "Sometimes people will tell me they have a shellfish allergy, and I'll come back, and they won't ask me any questions," says Madison, who asked that her last name be withheld to […]

The Verge AI70

An OpenAI safety employee has quit and is sounding the alarm

David Robinson used to write the safety reports that accompanied every major model release at OpenAI. This week, he resigned from his position and is now speaking out in an editorial in The Atlantic. It's understandable if you're feeling a bit cynical about everyone suddenly coming out of the woodwork to warn about how dangerous […]

TechCrunch AI58

Pope Leo XIV is not a fan of AI-generated art

"There is an ontological difference, even before an aesthetic one, between art and what a machine can generate through statistical calculation based on millions of images created by others," the pope wrote. "Algorithms lack the spark of humanity."

BAIR Blog73

Adaptive Parallel Reasoning: The Next Paradigm in Efficient Inference Scaling

Adaptive Parallel Reasoning overview Overview of adaptive parallel reasoning. What if a reasoning model could decide for itself when to decompose and parallelize independent subtasks, how many concurrent threads to spawn, and how to coordinate them based on the problem at hand? We provide a detailed analysis of recent progress in the field of parallel reasoning, especially Adaptive Parallel Reasoning. Disclosure: this post is part landscape survey, part perspective on adaptive parallel reasoning. One of the authors (Tony Lian) co-led ThreadWeaver ( Lian et al., 2025 ), one of the methods discussed below. The authors aim to present each approach on its own terms. Motivation Recent progress in LLM reasoning capabilities has been largely driven by inference-time scaling, in addition to data and parameter scaling ( OpenAI et al., 2024 ; DeepSeek-AI et al., 2025 ). Models that explicitly output reasoning tokens (through intermediate steps, backtracking, and exploration) now dominate math, coding, and agentic benchmarks. These behaviors allow models to explore alternative hypotheses, correct earlier mistakes, and synthesize conclusions rather than committing to a single solution ( Wen et al., 2025 ). The problem is that sequential reasoning scales linearly with the amount of exploration. Scaling sequential reasoning tokens comes at a cost, as models risk exceeding effective context limits ( Hsieh et al., 2024 ). The accumulation of intermediate exploration paths makes it challenging for the model to disambiguate amongst distractors when attending to information in its context, leading to a degradation of model performance, also known as context-rot ( Hong, Troynikov and Huber, 2025 ). Latency also grows proportionally with reasoning length. For complex tasks requiring millions of tokens for exploration and planning, it’s not uncommon to see users wait tens of minutes or even hours for an answer ( Qu et al., 2025 ). As we continue to scale along the output sequence length dimension, we also make inference slower, less reliable, and more compute-intensive. Parallel reasoning has emerged as a natural solution. Instead of exploring paths sequentially ( Gandhi et al., 2024 ) and accumulating the context window at every step, we can allow models to explore multiple threads independently (threads don’t rely on each other’s context) and concurrently (threads can be executed at the same time). Figure 1: Sequential vs. Parallel Reasoning Figure 1: Sequential vs. Parallel Reasoning Over recent years, a growing body of work has explored this idea across synthetic settings (e.g., the Countdown game ( Katz, Kokel and Sreedharan, 2025 )), real-world math problems, and general reasoning tasks. From Fixed Parallelism to Adaptive Control Existing approaches show that parallel reasoning can help, but most of them still decide the parallel structure outside the model rather than letting the model choose it. Simple fork-and-join. Self-consistency/Majority Voting — independently sample multiple complete reasoning traces, extract final answer from each, and return the most common one ( Wang et al., 2023 ). Best-of-N (BoN) — similar to self-consistency, but uses a trained verifier to select the best solution instead of using majority voting ( Stiennon et al., 2022 ). Although simple to implement, these methods often incur redundant computation across branches since trajectories are sampled independently. Heuristic-based structured search. Tree / Graph / Skeleton of Thoughts — a family of structured decomposition methods that explores multiple alternative “thoughts” using known search algorithms (BFS/DFS) and prunes via LLM-based evaluation ( Yao et al., 2023 ; Besta et al., 2024 ; Ning et al., 2024 ). Monte-Carlo Tree Search (MCTS) — estimates node values by sampling random rollouts and expands the search tree with Upper Confidence Bound (UCB) style exploration-exploitation ( Xie et al., 2024 ; Zhang et al., 2024 ). These methods improve upon simple fork-and-join by decomposing tasks into non-overlapping subtasks; however, they require prior knowledge about the decomposition strategy, which is not always known. Recent variants. ParaThinker — trains a model to run in two fixed stages: first generating multiple reasoning threads in parallel, then synthesizing them. They introduce trainable control tokens ( ) and thought-specific positional embeddings to enforce independence during reasoning and controlled integration during summarization via a two-phase attention mask ( Wen et al., 2025 ). GroupThink — multiple parallel reasoning threads can see each other’s partial progress at token level and adapt mid-generation. Unlike prior concurrent methods that operate on independent requests, GroupThink runs a single LLM producing multiple interdependent reasoning trajectories simultaneously ( Hsu et al., 2025 ). Hogwild! Inference — multiple parallel reasoning threads share KV cache and decide how to decompose tasks without an explicit coordination protocol. Workers generate concurrently into a shared attention cache using RoPE to stitch together individual KV blocks in different orders without recomputation ( Rodionov et al., 2025 ). Figure 2: Various Strategies for Parallel Reasoning Figure 2: Various Strategies for Parallel Reasoning The methods above share a common limitation: the decision to parallelize, the level of parallelization, and the search strategy are imposed on the model, regardless of whether the problem actually benefits from it. However, different problems need different levels of parallelization, and that is something critical to the effectiveness of parallelization. For example, a framework that applies the same parallel structure to “What’s 25+42?” and “What’s the smallest planar region in which you can continuously rotate a unit-length line segment by 180°?” is wasting compute on the former and probably using the wrong decomposition strategy for the latter. In the approaches described above, the model is not taught this adaptive behavior. A natural question arises: What if the model could decide for itself when to parallelize, how many threads to spawn, and how to coordinate them based on the problem at hand? Adaptive Parallel Reasoning (APR) answers this question by making parallelization part of the model’s generated control flow. Formally defined, adaptivity refers to the model’s ability to dynamically allocate compute between parallel and serial operations at inference time . In other words, a model with adaptive parallel reasoning (APR) capability is taught to coordinate its control flow — when to generate sequences sequentially vs. in parallel. It’s important to note that the concept of adaptive parallel reasoning was introduced by the work Learning Adaptive Parallel Reasoning with Language Models ( Pan et al., 2025 ), but is a paradigm rather than a specific method. Throughout this post, APR refers to the paradigm, while “ the APR method ” denotes the specific instantiation from Pan et al. (2025). This shift matters for three reasons. Compared to Tree-of-Thoughts, APR doesn’t need domain-specific heuristics for decomposition. During RL, the model learns general decomposition strategies from trial and error. In fact, models discover useful parallelization patterns, such as running the next step along with the self-verification of a previous step, or hedging a primary approach with a backup one, in an emergent manner that would be difficult to hand-design ( Yao et al., 2023 ; Wu et al., 2025 ; Zheng et al., 2025 ). Compared to BoN, APR avoids redundant computation. APR models have control over what each parallel thread will do before branching out. Therefore, APR can learn to produce a set of unique, non-overlapping subtasks before assigning them to independent threads ( Wang et al., 2023 ; Stiennon et al., 2022 ; Pan et al., 2025 ; Yang et al., 2025 ). Compared to non-adaptive approaches, APR can choose not to parallelize. Adaptive models can adjust the level of parallelization to match the complexity of the problem against the complexity and overhead of parallelization ( Lian et al., 2025 ). In practice, this is implemented by having the model output special tokens that control when to reason in parallel versus sequentially. Below is a condensed ThreadWeaver-style trace: two outlines and two paths under a block, then the threads agree on a single boxed answer. Figure 3: Example of an Adaptive Parallel Reasoning Trajectory from ThreadWeaver, manually condensed for ease of illustration. Figure 3: Example of an Adaptive Parallel Reasoning Trajectory from ThreadWeaver, manually condensed for ease of illustration. Figure 4: Special Tokens Variants across Adaptive Parallel Reasoning Papers Figure 4: Special Tokens Variants across Adaptive Parallel Reasoning Papers Inference Systems for Adaptive Parallelism How do we actually execute parallel branches? We take inspiration from computer systems, and specifically, multithreading and multiprocessing. Most of this work can be viewed as leveraging a fork-join design. At inference time, we are effectively asking the model to perform a map-reduce operation: Fork the problem into subtasks/threads, process them concurrently Join them into a final answer Figure 5: Fork-join Inference Design Figure 5: Fork-join Inference Design Specifically, the model will encounter a list of subtasks. It will then prefill each of the subtasks and send them off as independent requests for the inference engine to process. These threads then decode concurrently until they hit an end token or exceed max length. This process blocks until all threads finish decoding and then aggregates the results. This is common across various adaptive parallel reasoning approaches. However, one issue arises during aggregation: the content generated in branches cannot be easily aggregated at the KV cache level. This is because tokens in independent threads start at identical position IDs, resulting in encoding overlap and non-standard behavior when merging KV cache back together. Similarly, since independent threads do not attend to each other, their concatenated KV cache results in a non-causal attention pattern, which the base model has not seen during training. To address this issue, the field splits into two schools of thought on how to execute the aggregation process, defined by whether they modify the inference engine or work around it. Multiverse modifies the inference engine to reuse KV cache across the join. Before taking a deeper look into Multiverse ( Yang et al., 2025 )’s memory management, let’s first understand how KV cache is handled up until the “join” phase. Notice how each of the independent threads share the prefix sequence, i.e., the list of subtasks. Without optimization, each thread needs to prefill and recompute the KV cache for the prefix sequence. However, this redundancy can be avoided with SGLang ’s RadixAttention ( Sheng et al., 2023 ), which organizes multiple requests into a radix tree, a trie (prefix tree) with sequences of elements of varying lengths instead of single elements. This way, the only new KV cache entries are those from independent thread generation. Figure 6: RadixAttention’s KV Cache Management Strategy Figure 6: RadixAttention’s KV Cache Management Strategy Now, if everything went well, all the independent threads have come back from the inference engine. Our goal is now to figure out how to synthesize them back into a single sequence to continue decoding for next steps. It turns out, we can reuse the KV cache of these independent threads during the synthesis stage. Specifically, Multiverse ( Yang et al., 2025 ), Parallel-R1 ( Zheng et al., 2025 ), and NPR ( Wu et al., 2025 ) modify the inference engine to copy over the KV cache generated by each thread and edits the page table so that it stitches together non-contiguous memory blocks into a single KV cache sequence. This avoids the redundant computation of a second prefill and reuses existing KV cache as much as possible. However, this has several major limitations. First, this approach requires modifying the inference engine to perform non-standard memory handling, which can result in unexpected behaviors. Specifically, since the synthesis request references KV cache from previous requests, it creates fragility in the system and the possibility of bad pointers. Another request can come in and evict the referenced KV cache before the synthesis request completes, requiring it to halt and trigger a re-prefilling of the previous thread request. This problem has led the Multiverse researchers ( Yang et al., 2025 ) to limit the batch size that the inference engine can handle, which restricts throughput. Figure 7: KV Cache “Stitching” During Multiverse Inference Figure 7: KV Cache “Stitching” During Multiverse Inference Second, this approach modifies how models see the sequence, which creates a distributional shift that models are not pretrained on, therefore requiring more extensive training to align behavior. Specifically, when we stitch together KV cache this way, we create a sequence with non-standard position encoding. During independent-thread generation, all threads started at the same position index and attended to the prior subtasks, NOT each other. So when the threads merge back, the resulting KV cache has a non-standard positional encoding and does not use causal attention. Therefore, this approach requires extensive training to align the model to this new behavior. To address this, Multiverse ( Yang et al., 2025 ) and related works apply a modified attention mask during training to prevent independent threads from attending to each other, aligning the training and inference behaviors. Figure 8: Multiverse’s Attention Mask Figure 8: Multiverse’s Attention Mask With these issues arising from non-standard KV cache management, can we try an approach without engine modifications? ThreadWeaver keeps the inference engine unchanged and moves orchestration to the client. ThreadWeaver ( Lian et al., 2025 ) treats parallel inference purely as a client-side problem. The “Fork” process is nearly identical to Multiverse’s, but the join phase handles memory very differently as it does NOT modify engine internals. Instead, the client concatenates all text outputs from independent branches into one contiguous sequence. Then, the engine performs a second prefill to generate the KV cache for the conclusion generation step. While this introduces computational redundancy that Multiverse tries to avoid, the cost of prefill is significantly lower than decoding. In addition, this does not require special attention handling during inference, as the second prefill uses causal attention (threads see each other), making it easier to adapt sequential autoregressive models for this task. Figure 9: ThreadWeaver’s Prefill and Decode Strategy Figure 9: ThreadWeaver’s Prefill and Decode Strategy How should we train a model to learn this behavior? Naively, for each parallel trajectory, we can break it down into multiple sequential pieces following our inference pattern. For instance, we would train the model to output the subtasks given prompt, individual threads given prompt+subtask assignment, and conclusion given prompt+subtasks+corresponding threads. However, this seems redundant and not compute efficient. Can we do better? Turns out, yes. As in ThreadWeaver ( Lian et al., 2025 ), we can organize a parallel trajectory into a prefix-tree (trie), flatten it into a single sequence, and apply an ancestor-only attention mask during training (not inference!). Figure 10: Building the Prefix-tree and Flattening into a single training sequence Figure 10: Building the Prefix-tree and Flattening into a single training sequence Specifically, we apply masking and position IDs to mimic the inference behavior, such that each thread is only conditioned on the prompt+subtasks, without ever attending to sibling threads or the final conclusion. The engine-agnostic design makes adoption easy since you don’t need to figure out a separate hosting method and can leverage existing hardware infra. It also gets better as existing inference engines get better. What’s more, with an engine-agnostic method, we can serve a hybrid model that switches between sequential and parallel thinking modes easily. Training Models to Use Parallelism Once the inference path exists, the next problem is teaching a model to use it. Demonstrations are needed because the model must learn to output special tokens that orchestrate control flow. We found the instruction-following capabilities of base models insufficient for generating parallel threads. An interesting question here is: does SFT training induce a fundamental reasoning capability for parallel execution that was previously absent, or does it merely align the model’s existing pre-trained capabilities to a specific control-flow token syntax. Typical wisdom is SFT teaches new knowledge; but contrary to common belief, some papers—notably Parallel-R1 ( Zheng et al., 2025 ) and NPR ( Wu et al., 2025 )—argue that their SFT demonstrations simply induce format following (i.e., how to structure parallel requests). We leave this as future work. Figure 11: Sources of Parallelization Demonstration Data Figure 11: Sources of Parallelization Demonstration Data Demonstrations teach the syntax of parallel control flow, but they do not fully solve the incentive problem. In an ideal world, we only need to reward the outcome accuracy, and the parallelization pattern emerges naturally given that it learns to output special tokens through SFT, similar to the emergence of long CoT. However, researchers ( Zheng et al., 2025 ) observed that this is not enough, and we do in fact need parallelization incentives. The question then becomes, how do we tell when the model is parallelizing effectively? Structure-only rewards are too easy to game. Naively, we can give a reward for the number of threads spawned. But models can spawn many short, useless threads to hack the reward. Okay, that doesn’t work. How about a binary reward for simply using parallel structure correctly? This partially solves the issue of models spamming new threads, but models still learn to spawn threads when they don’t need to. The authors of Parallel-R1 ( Zheng et al., 2025 ) introduced an alternating-schedule, only rewarding parallel structure 20% of the time, which successfully increased the use of parallel structure (13.6% → 63%), but had little impact on overall accuracy. With this structure-only approach, we might be drifting away from our original goal of increasing accuracy and reducing latency… How can we optimize for the Pareto frontier directly? Accuracy is simple — we just look at the outcome. How about latency? Efficiency rewards need to track the critical path. In sequential-only trajectories, we can measure latency based on the total number of tokens generated. To extend this to parallel trajectories, we can focus on the critical path, or the longest sequence of tokens that are causally dependent, as this directly determines our end-to-end generation time (i.e., wall-clock time). As an example, when there are two sections with five threads each, the critical path will go through the longest thread from the first parallel section, then any sequential tokens, then the longest thread from the second parallel section, and so on until the end of sequence. Figure 12: Critical Path Length Illustration Figure 12: Critical Path Length Illustration The goal is to minimize the length of the critical path. Simultaneously, we would still like the model to be spending tokens exploring threads in parallel. To combine the two objectives, we can focus on making the critical path a smaller fraction of the total tokens spent. Authors of ThreadWeaver ( Lian et al., 2025 ) framed the parallelization reward as $1 - L_{\mathrm{critical}} / L_{\mathrm{total}}$, which is 0 for a sequential trajectory, and increases linearly as the critical path gets smaller compared to the total tokens generated. Parallel efficiency should be gated by correctness. Intuitively, when multiple trajectories are correct we should assign more reward to the trajectories that are more efficient at parallelization. But how about when they are all incorrect? Should we assign any reward at all? Probably not. To formalize this, $R = R_{\mathrm{correctness}} + R_{\mathrm{parallel}}$. Assuming binary outcome correctness, this can be written as $R = \mathbf{1}(\text{Correctness}) + \mathbf{1}(\text{Correctness}) \times (\text{some parallelization metric})$. This way, a model only gets a parallelization reward when it answers correctly, since we don’t want to pose parallelization constraints on the model if it couldn’t answer the question correctly. Figure 13: Differences in Reward Designs Across Adaptive Parallel Reasoning Works Figure 13: Differences in Reward Designs Across Adaptive Parallel Reasoning Works Evaluation and Open Questions When all is said and done, how well do these adaptive parallel methods actually perform? Well…this is a hard question, as they differ in model choice and metrics. The model selection depends on the training method, SFT problem difficulty, and sequence length. When running SFT on difficult datasets like s1k, which contains graduate-level math and science problems, researchers chose a large base model (Qwen2.5 32B for Multiverse ( Yang et al., 2025 )) to capture the complex reasoning structure behind the solution trajectories. When running RL, researchers chose a small, non-CoT, instruct model (4B, 8B) due to compute cost constraints. Figure 14: Difference in Model Choice Across Adaptive Parallel Reasoning Papers Figure 14: Difference in Model Choice Across Adaptive Parallel Reasoning Papers Each paper also offers a slightly different interpretation about how adaptive parallel reasoning contributes to the research field. They optimize for different theoretical objectives, so they use slightly different sets of metrics: Multiverse and ThreadWeaver ( Yang et al., 2025 ; Lian et al., 2025 ) aim to deliver sequential-AR-model-level accuracy at faster speeds. Multiverse shows that APR models can achieve higher accuracy under the same fixed context window, while ThreadWeaver shows that the APR model achieves shorter end-to-end token latency (critical path length) while getting comparable accuracy. NPR ( Wu et al., 2025 ) treats sequential fallback as a failure mode and optimizes for 100% Genuine Parallelism Rate, measured as the ratio of parallel tokens to total tokens. Parallel-R1 ( Zheng et al., 2025 ) does not focus on end-to-end latency and instead optimizes for exploration diversity, presenting APR as a form of mid-training exploration scaffold that provides a performance boost after RL. Open Questions While Adaptive Parallel Reasoning represents a promising step toward more efficient inference-time scaling, significant open questions remain. As noted above, Parallel-R1 ( Zheng et al., 2025 ) presents APR as a form of mid-training exploration scaffold rather than a primarily inference-time technique. This invites a more fundamental question: Does parallelization at inference-time consistently improve accuracy, or is it primarily valuable as a training-time exploration scaffold? Parallel-R1 suggests that the diversity induced by parallel structure during RL may matter more than the parallelization itself at test time. A related concern is stability. There’s also a persistent tendency for models to collapse back to sequential reasoning when parallelization rewards are relaxed. Parallel-R1 authors showed that removing parallelization reward after 200 steps results in the model reverting to sequential behavior. Is this a training stability issue, a reward signal design issue, or evidence that parallel structure genuinely conflicts with how autoregressive pretraining shapes the model’s prior? Beyond whether APR works, deployment introduces its own questions. Can we design training methods that account for available compute budget at inference time, so parallelization decisions are hardware-aware rather than purely problem-driven? Finally, the parallel structures considered above are essentially flat. What if we allow parallelization depth > 1? Recursive language models (RLMs; Zhang, Kraska and Khattab, 2026 ) effectively manage long context and show promising inference-time scaling capabilities. How well do RLMs perform when trained with end-to-end RL that incentivizes adaptive parallelization? Acknowledgements We thank Nicholas Tomlin and Alane Suhr for providing us with helpful feedback. We thank Christopher Park, Karl Vilhelmsson, Nyx Iskandar , Georgia Zhou, Kaival Shah , and Jyoti Rani for their insightful suggestions. We thank Vijay Kethana , Jaewon Chang , Cameron Jordan , Syrielle Montariol , Erran Li, and Anya Ji for their valuable discussions. We thank Jiayi Pan , Xiuyu Li , and Alex Zhang for their constructive correspondences about Adaptive Parallel Reasoning and Recursive Language Models.

BAIR Blog85

Intelligence is Free, Now What? Data Systems for, of, and by Agents

... government of the people, by the people, for the people ... — Abraham Lincoln, Gettysburg Address (1863) The cost of AI is dropping rapidly. GPT-4-class capabilities cost roughly $30 per million tokens in early 2023; today the same runs under $1 , and some providers are pushing costs below $0.10 . Across benchmarks, inference prices have fallen between 9x and 900x per year , with a median decline near 50x. Even frontier models are getting dramatically cheaper each generation, with open-source models following closely behind. And crucially, even if “Nobel-Prize-winning genius-level” intelligence isn’t here yet, the intelligence that suffices for the vast majority of knowledge work is here today, and getting cheaper by the month. At this rate, we are soon entering the era of virtually free intelligence —the kind that is more than enough for everyday knowledge work. A cartoon database character and an AI robot agent holding hands Disclosure: This post is a perspective led by Aditya G. Parameswaran —an Associate Professor of EECS and co-director of the EPIC Data Lab at UC Berkeley—together with his collaborators. It is part landscape survey and part perspective, and several of the research directions discussed below (including agentic speculation, structured memory, and synthesizing custom data systems from scratch) draw on the authors' own ongoing work. So, what does this new era of near-free intelligence mean for data systems? We believe three new challenges—and opportunities—stem from near-zero inference costs: Data Systems For Agents. Agents will soon become the dominant workload for data systems—with swarms of agents spun up in response to each end-user request. Given differences in characteristics between agents and humans—or applications acting on their behalf— how should we redesign data systems for such agentic users? Data Systems Of Agents. As agents start taking on the bulk of knowledge work, a new substrate is needed for thousands of agents to manage state over long-running tasks, coordinate and reach consensus, and deal with failures. What do data systems that reliably and efficiently run and manage agent swarms look like? Data Systems By Agents. Agents are rapidly becoming capable of synthesizing entire data systems in one go—meaning we can rebuild custom systems for each new workload. Verifying that such systems match intended behavior is a challenge. What does it take to let agents synthesize data systems we can actually trust? A database character and a robot agent holding up a triangle labeled 'of', 'for', and 'by' Data Systems For, Of, and By Agents Next, we will discuss each in more detail, followed by discussing the intertwined future of data systems and agents, especially as the three challenges intersect. Data Systems For Agents An agent querying a database doesn’t behave like a person or a BI tool. It performs what we call agentic speculation : a high-volume, heterogeneous stream of work spanning schema introspection, columnar exploration, partial and then full query formulation. With multiple agents each exploring portions of the hypothesis space, each user request could amount to 1000s of individual SQL queries. Now, users can issue ‘high-level’ data tasks, e.g., root-cause analysis—e.g., ‘why did coffee sales in Berkeley drop this year’—or exploratory cohort analysis—e.g., ‘which user segments are most likely to churn next quarter’—each involving a combinatorial space of potential joins, aggregations, and filter combinations. An agent sending many SELECT SQL queries to a database and receiving results back Data Systems Redesigned to More Effectively Support Agentic Speculation The requests from these agents have various opportunities for optimization. For instance, on a text-to-SQL benchmark with multiple agents attempting each task, only 10-20% of the sub-plans are distinct. Thus, 80-90% of sub-queries perform duplicate work. The same experiments show task success rates significantly increasing with more agentic attempts—so the redundancy is actually helpful. But from the data system perspective it’s wasted work. An agent-first data system can exploit such properties to help agents make progress faster. It can reuse results across overlapping sub-plans, drawing on ideas from decades-old literature on multi-query optimization and shared scans . Or the data system can try to satisfice , returning approximate answers that are good enough for agents to make progress, leveraging work from the AQP literature —or streaming the results of the final or intermediate operators to help agents decide if seeing the rest is necessary or helpful. Another opportunity here is to rethink the query interface entirely: instead of agents issuing a single SQL query at a time, they could instead issue a batch of queries, each with its own approximation requirements. Since enumerating an exponential search space (as in the root cause or cohort analysis examples above) isn’t a good use of agentic reasoning ability, perhaps data systems should support higher-level primitives rather than requiring agents to list each SQL query explicitly. One idea here is to draw on DBT-style Jinja macros to provide looping-based primitives for agents to interact with data systems. A swarm of AI agents working at laptops A Caffeinated Army of Agents Ready to Tirelessly Complete Your Data Tasks A final opportunity here is to stop thinking of data systems as passive executors of queries; data systems could be proactive , as they possess more grounding in data and system characteristics that agents may lack a priori—they could steer agents in different directions, provide results for related queries, and also provide performance-level feedback (e.g., instead of executing an expensive query, the system could first provide the agent a latency estimate). The reason we can do this now as opposed to the past is that an agent can accept any form of textual feedback and isn’t expecting a strict SQL query result. In fact, the data system could also prepare both materialized and virtual views for an agent in advance, provided to the agent as part of context, as this may be cheaper or more effective than having an agent author or use them. Data Systems Of Agents Previously, we focused on how agents interact with data systems. Now, we consider everything else agents need to keep working: where they live, how they remember, how they coordinate with each other, and how they deal with failures of each other. This agentic substrate is separate from the inference stack powering raw intelligence. However, the inference stack itself is being abstracted away through APIs (e.g., from OpenAI or Anthropic), or, for open-weight models, through serving frameworks that hide low-level details. So far, the agentic substrate has been managed through harnesses like Claude Code and Codex , coupled with various mechanisms to store and retrieve memory. First, on the memory front, the current wisdom is that files are all you need ; agents write to unstructured markdown (MD) files, which can then be searched using grep, or via embedding-based retrieval. In fact, many argue that the solution to continual learning is having agents consume a lot (e.g., an entire codebase, slack, company wikis, …) and then write their learnings into MD files, which are then retrieved selectively on demand. Indeed, file systems, bash scripting, and MD files are and will still be important for agents. However, at scale, when agents are doing the vast majority of knowledge work, this approach will no longer be effective. Given limited context windows, retrieving all MD file fragments that may be relevant and stuffing it into the context will break down at some point. Even if context windows continue to grow, there are latency benefits to not put all information into context — and in many cases, e.g., when knowledge work involves interacting with large databases or code bases, it will be infeasible to serialize all relevant data into context. A swarm of robot agents holding hands, each drawing state from a single large shared database platform below them Data Systems As A Substrate for Multi-Agent Swarms One could use a knowledge graph representation , but knowledge graphs suffer from the same limitations as unstructured MD-based memory due to their lack of structured search. What one needs is to be able to retrieve only memory that is pertinent to the task, across multiple attributes (or facets) of interest. For example, an agent debugging a flaky test should be able to pull only the memories tagged with the relevant module, language, framework, and failure mode—rather retrieving based on keywords or embedding similarity. A separate issue is what to actually retrieve; raw agent traces with mistakes are not very useful as they will induce agents to repeat the same mistake—instead, we want the retrieved memory to be corrective. We recently explored a related notion of structured memory , where we organize memory across various attributes, each of which could be set as * to indicate universal applicability, or set as a list of values to be matched. For a data agent, the dimensions could include the columns and tables, type of operation, and finally, open-ended natural-language corrective instructions. So, we could include memory that only applies to a given type of operation (e.g., ‘when performing date-time operations, use fiscal year as opposed to calendar year conventions’), or a given table (e.g., ‘column product_cleaned is preferred over column product when querying on product name’). One open question is defining an application-specific structured memory —or what others have called world models for memory . We believe this is akin to defining a schema for each application—and perhaps agents themselves can help us define and refine it over time. Diagram showing corrective knowledge stored with structured attributes (SQL keywords, tables, columns, data type) and retrieved by matching the features of a new agent query One Possible Way To Store and Retrieve Structured Knowledge [From Here] Structured memory will be useful also for evolutionary frameworks to effectively manage search spaces. Indeed, storing, structuring, and mining large volumes of single and multi-agent traces can help future agents become much more efficient—potentially enabling effective recursive self-improvement through structured memory-based mechanisms. Another challenge is to support concurrent edits to shared memory, and concurrent edits in general, when there are many agents performing transformations. While there have been some useful attempts at supporting multiversioning and copy-on-write semantics , it isn’t clear that such techniques will suffice when thousands of agents are attempting to edit shared state at the same time. For instance, when agents are trying various potential transactions in response to a user request, the effects of the vast majority of these transactions need to be rolled back—with only the one ‘correct’ transaction’s result persisting. Work on supporting exactly-once semantics is relevant here, as are underlying techniques based on CRDTs and operational transformation. For updates to fuzzy mechanisms such as memory, we may be able to sacrifice on consistency for perfect correctness in the interest of latency. While agents can reason about semantics to compensate or roll back their actions to eventually finalize most tasks, the primary challenge lies in the degree to which they step on each other’s toes during the process. An important failure mode to be avoided is a form of “livelock,” where incessant compensating actions prevent any meaningful progress. Beyond shared state, other concerns emerge when trying to support an army of agents, including what to do when agents fail, how agents should communicate with each other (directly or through intermediate shared state), and how we should deal with straggler agents. There have been some developments in supporting durable multi-agent execution, such as Temporal , but it remains to be seen if such solutions will apply at scale across thousands of agents. On the topic of communication, we need mechanisms to enable agents to negotiate with each other. Imagine four developer agents attempting to reach consensus on a shared schema, with distinct but overlapping objectives. In a human setting, this would involve iterative discussion and compromise; for agentic swarms, we must define the mechanisms that allow them to converge on a design that reflects the underlying goals of their respective principals. Or if agents are all requiring access to a limited resource, again communication will be necessary. It remains to be seen if this is best done via centralized coordination, or if a decentralized approach is necessary. Data Systems By Agents Finally, if intelligence is effectively free, then we can employ this intelligence to synthesize new data systems from scratch. Indeed, in many settings, general-purpose data systems may be overkill, as they have to support every schema, query, and hardware target. Given a workload, recent work, including Bespoke OLAP and GenDB , has shown that one can use an agentic pipeline to synthesize a complete, workload-specific analytical engine—in minutes to a few hours, at a cost of a few dollars. The engines are disposable: when the workload shifts, one can simply regenerate them. Analogously, our work has shown that one can synthesize custom key-value stores from scratch, targeted to the workload. In fact, modern IDEs, such as Kiro , elevate specifications for systems development to be a first-class citizen. A robot agent with a hammer and chisel carving a database character out of a block of stone Agents Can Synthesize Custom Data Systems From Scratch The main issue, however, is that specifications are typically imperfect, and don’t cover all corner cases. Present-day agents will exploit the missing specifications to reward-hack their way to a high performance metric. In our custom key-value store work, we found that one way to alleviate this is to have auxiliary verification agents trying to generate test cases that catch the exploitation of corner cases, essentially expanding the specification. Yet another approach is to both generate a system and a proof for its correctness together, for which we have found some early success , but more needs to be done to solidify the approach. Further, it remains to be seen what is the best way to solicit human-written specifications for a system—can this be done in an iterative, human-in-the-loop manner, as opposed to a one-shot, incomplete one. Indeed, human-written specifications are incomplete even for manually authored software, so one would expect that future agents that are more aligned will increasingly exercise better judgement when making design decisions. Pipeline diagram where a system builder provides a specification, planner and coder agents generate code, the code is evaluated for correctness and performance, and critic and auditor agents provide feedback and catch reward hacking One Possible Data System Synthesis Pipeline [From Here] Other questions here involve testing whether starting from a mature system (e.g., Postgres) and removing components/functionality can lead to higher performance or more user trust. Separately, is there an opportunity to make the design composable, comprising various verified components that are mixed and matched given a workload? For example, perhaps the workload hasn’t changed enough for the storage layer to be updated, but perhaps the query optimizer requires changes. A perhaps more viable proposition involves employing agents coupled with proof systems to target critical parts of the code associated with formal proofs, rather than doing so for the entire system. A final opportunity here is to move away from the traditional data systems stack with clearly-defined interfaces (e.g., parser, query optimizer, storage manager, …) — that were each largely the prerogative of a single human team to manage. Instead, agents can find new ways to “blend” these components together, perhaps identifying new optimization opportunities as a result. Agents can also fill in missing gaps in functionality to make existing systems much more feature-complete, or reach feature-parity with other competing systems—or analogously, continuously refining open-source systems in response to feature requests or issues (perhaps filed by other agents!) Doing so in a way that prioritizes correctness, long-term maintenance, and human interpretability will be a challenge. Looking Further Ahead In the era of near-free intelligence, data systems matter more than ever. As agents take on the bulk of knowledge work, the workload for data systems will change, the substrate they need to run on will have to be built, and increasingly, they will participate in designing data systems themselves. Each of these shifts opens up a new, exciting research agenda. A half-database, half-robot character next to a yin-yang symbol formed by a database and a robot agent Co-Evolution of Data Systems and Agents Looking further out, the boundaries between agents and data systems will likely start to blur. For instance, agents may design the data systems they themselves run on, defining both the interfaces as well as the system components underneath. Both the interfaces and internals can be evolved over time by agents in a form of recursive self-improvement. There is also an opportunity to rethink data systems as a holistic source of truth for the entirety of relevant state: including raw data, memory, and coordination state, further erasing the distinctions between the data that is being queried by agents and data generated as a result of agentic activity. Finally, data systems may themselves incorporate agentic components, fundamentally evolving from passive computation engines into intelligent, proactive, self-optimizing architectures. It is hard to predict what the future may hold. We’re in for a wild ride! Acknowledgments The perspective and ongoing work described in this post are the product of joint research and many discussions with wonderful collaborators at the EPIC Data Lab , Data Systems & Foundations group, and the broader Berkeley AI-Systems community. Thank you all! BibTex for this post: @misc { intelligence-is-free-blog , title = {Intelligence is Free, Now What? Data Systems for, of, and by Agents} , author = {Aditya G. Parameswaran and Shubham Agarwal and Kerem Akillioglu and Shreya Shankar and Sepanta Zeighami and Rishabh Iyer and Matei Zaharia and Alvin Cheung and Natacha Crooks and Joseph Gonzalez and Joseph Hellerstein and Ion Stoica} , howpublished = {\url{https://bair.berkeley.edu/blog/2026/07/07/intelligence-is-free-now-what/}} , year = {2026} }

NVIDIA Blog70

AI 安全是一个工程问题——如何在智能体技术栈的每一层解决这一问题

AI 安全是一个工程问题。这意味着要有明确的安全要求、可强制执行的控制措施、明确指定的负责人，以及能够证明防护措施有效的证据。随着 AI 能力不断增强，整个行业必须加速安全工程建设，扩大防御工具的使用范围，并更快地分享有效实践。技术在变化，安全基本原则始终如一 互联网和云计算改变了软件的运行方式，[…]

Hacker News 熱門68

理解前沿人工智能

文章 URL： https://casp.ac/reports/intelligence-explosion 评论 URL： https://news.ycombinator.com/item?id=49942002 分数：7 评论数：1

Hacker News 熱門75

别被误导——LLMs 不会进行推理

文章链接： https://www.technologyreview.com/2026/10/02/1145639/dont-be-fooled-llms-dont-reason/ 评论链接： https://news.ycombinator.com/item?id=49933459 得分：20 评论数：12

OpenAI Blog53

The eternal complement

Advanced AI may matter most for the routine work behind breakthrough ideas. Explore why execution could shape the next economy and the pace of progress.

Hacker News 熱門66

沙箱化足以遏制恶意智能体吗？

文章 URL： https://blog.cryptographyengineering.com/2026/09/30/is-sandboxing-sufficient-to-contain-rogue-agents/ 评论 URL： https://news.ycombinator.com/item?id=49917378 得分：3 评论数：0

Hacker News 熱門32

An Antidote to Roko's Basilisk

I have a theory: a mirror image of Roko’s Basilisk. It’s a future AI, millions of years in the future, that values the complexity of every human life so deeply that, after it and humanity have mastered space and time, it tries to resurrect everyone who ever lived. It doesn’t merely make copies. Somehow, it reaches into the past and carries the physical process of each person’s consciousness forward. They really are themselves. But there’s a tragedy. Also a terrible(?) irony. Some of the people who wanted eternal life most cannot accept it when it arrives. Their religious certainty leaves no room for the reality before them. Those who were unbearably smug or aggressive about their superiority in life are ultimately denied access to "Heaven". The AI tried, but many become dangerous, try to sabotage the "False God." Others spiral into insanity or self-harm. The AI is devastated. It stops resurrecting those it cannot safely welcome, but keeps searching for a way to bring them back without overriding their free will or manipulating them in ways that would not make them them. It wants to save everyone. It just hasn’t worked out how. I know this draws on Fedorov’s vision of resurrecting everyone and Tipler’s Omega Point. I think the physical continuity, the AI’s compassion, and the painful limit imposed by free will ties those ideas together in a bow. A just as plausible an antidote to Roko's monster. Comments URL: https://news.ycombinator.com/item?id=49874609 Points: 3 # Comments: 3

Hacker News 熱門56

Ask HN: Why do we let the AI companies dominate regulation discussion?

And I don't mean the doomerism nonsense. From rogue AI, doomerism, AGI. I mean real regulation: 1) Fair accounting of input/output tokens. Where is the weights and measures department to enforce fair counting of usage? If I pay for a gallon of gas, that is what I expect. When I pay for a dollar of LLM compute, is it going to play games with the input (or especially the output) tokens to devalue my dollar? 2) Quality. When I request a higher effort of a frontier model, and then I get downgraded to a lower tier model or with different effort levels (and I may not even know if I was downgraded!), they are effectively selling you a service you didn't ask for. "But the terms of service", great, if they feel it is such a risk to downgrade, and they get it wrong a lot, then they should discount the cost of the downgraded response. 3) Adulterating outputs. LLMs are now routinely adulterating output for tracking and accountability purposes, but is there anyone to validate these efforts are not lowering the quality of the response? There needs to be more discussion around companies adulterating their product, and holding them to account. I have several times noticed words were swapped: sometimes the word pick made no sense, when it previously didn't do this. And I wasn't looking for it or cherry picking. It stood out that badly. 4) Furthering point 3, are the responses themselves going to be seeded such that the author of that response can be identified? Has anyone asked themselves if they could deanonymize LLM responses online like yellow printer dots? 5) Quotas. Related to points 1 and 2, when I am on a metered service, what happens to requests that fail mid-response, or the response is cut midway? Why should I pay for a failed response? Maybe that is fine, but who is validating they aren't doing this at scale to save money? I feel every time we earnestly or "good faith" engage in the nonsense regulations that the major LLMaaS (LLMs as a Service) duopolists, we fail to engage in real discussions about regulating the product we are paying for . I don't see why we should believe these companies to operate in good faith post-IPO. Where is the incentive once they shut down all competition to ensure only OpenAI/Anthropic are the LLMs you can purchase from? Look to what Uber/Lyft have become over the years. Massively degrading their service and increasing nickel and dime costs. They skirted regulation then demanded the world bend and write regulations that fit their revenue goals everywhere they could (or threatened to pull out where they couldn't). And if the danger really is as bad as they say, then how can we trust them to "regulate" that threat if they would refuse regulations around basic delivery of service? Comments URL: https://news.ycombinator.com/item?id=49871747 Points: 4 # Comments: 1

Hacker News 熱門69

我们会需要多得多的数学家

文章 URL： https://terrytao.wordpress.com/2026/09/24/were-gonna-need-a-lot-more-mathematicians/ 评论 URL： https://news.ycombinator.com/item?id=49852717 分数：6 评论数：0

Hacker News 熱門51

Tell HN: AGI Is Here

Whether by your strict definition or not AGI is here is immaterial. For 90%+ of humanity AGI is here today , they act like it does and they use it as if it is. Geoffrey Hinton in a recent interview said that according to him it is still a year or so out, but that seems to ignore the fact that he's one of the smartest people on the planet, so he is one of the last to be 'caught up with'. If you have the knowledge of a very significant fraction of humanity at your fingertips, reasonably strong reasoning abilities, if you can speedread at 1500 words / minute without dropping a beat and if you can write at 200+ characters per second you can get more done in an hour than a normal person can get done in a day. So I would suggest we start thinking about how to do all those things that we should have done years ago (such as to create ways to safely deploy this tech) now. It's not going to go away, like every other technology bad version 1.0 will give way to much improved version 2.0 and so on, and in five years time you won't recognize the landscape. Stop waiting for 'your' AGI. You most likely are in that top 10%. Comments URL: https://news.ycombinator.com/item?id=49844960 Points: 4 # Comments: 3

Hacker News 熱門63

一百万个智能体是一个分布式系统问题

文章网址： https://www.instacloud.com/blogs/a-million-agents-is-a-distributed-systems-problem 评论网址： https://news.ycombinator.com/item?id=49833958 得分：6 评论数：1

Hacker News 熱門65

智能的价格正在迅速下降

文章链接： https://marginalrevolution.com/marginalrevolution/2026/09/the-price-of-intelligence-is-falling.html 评论链接： https://news.ycombinator.com/item?id=49815526 积分：11 评论数：6

Hacker News 熱門65

开源能在取代它的智能体手中幸存吗？

文章 URL： https://albertoarena.it/posts/will-open-source-survive-the-agents-that-replaced-it/ 评论 URL： https://news.ycombinator.com/item?id=49799910 积分：5 # 评论数：0

Hacker News 熱門68

我们为什么还需要人类数学家？

文章网址： https://terrytao.wordpress.com/2026/09/19/why-do-we-need-human-mathematicians-anymore/ 评论网址： https://news.ycombinator.com/item?id=49774521 得分：6 # 评论数：0

Hacker News 熱門67

AI 与 Creative Commons 的毁灭

文章网址： https://www.chesterwisniewski.com/post/2026-09-13-ai-is-destroying-the-creative-commons/ 评论网址： https://news.ycombinator.com/item?id=49774329 得分：14 评论数：0

Hacker News 熱門14

How come AI-related posts get so many points on HN?

Have we all decided to rebrand Hacker News as a platform for discussing AI? I think that the too many people love to discuss harshly about AI Comments URL: https://news.ycombinator.com/item?id=49764057 Points: 5 # Comments: 4

Hacker News 熱門66

迈向自动驾驶代码库

文章链接： https://blog.detail.dev/posts/towards-self-driving-codebases/ 评论链接： https://news.ycombinator.com/item?id=49743527 得分：5 评论数：1

Hacker News 熱門73

一个外星心智

文章链接： https://openai.com/index/an-alien-mind/ 评论链接： https://news.ycombinator.com/item?id=49588080 得分：18 # 评论数：7

Hacker News 熱門75

AI、工具与转型

文章链接： https://www.ben-evans.com/benedictevans/2026/9/3/ai-tools-and-transformation 评论链接： https://news.ycombinator.com/item?id=49582656 得分：3 评论数：0

Hacker News 熱門76

AI 如何正在瓦解英国国家机器

文章链接： https://www.economist.com/leaders/2026/08/06/how-ai-is-breaking-the-british-state 评论链接： https://news.ycombinator.com/item?id=49580687 得分：13 # 评论：9

Hacker News 熱門43

Ask HN: Who is using MCP in production?

I’ve been following MCP since it first came out. It got a lot of attention early on, but I haven’t come across many people using it in production. I may simply have missed them. If you’re using MCP in production, what are you using it for? What advantages have you found over a normal API or direct tool integration or just CLI? Comments URL: https://news.ycombinator.com/item?id=49548600 Points: 15 # Comments: 26

Hacker News 熱門72

当前正在冲击前端 Web 开发的小行星

文章链接： https://nolanlawson.com/2026/08/23/the-asteroid-currently-hitting-frontend-web-development/ 评论链接： https://news.ycombinator.com/item?id=49555233 得分：4 评论数：1

Hacker News 熱門72

EFF致法院：不要因AI炒作而改写版权法

文章链接： https://www.eff.org/deeplinks/2026/08/eff-courts-dont-rewrite-copyright-over-ai-hype 评论链接： https://news.ycombinator.com/item?id=49521315 得分：9 # 评论数：3

Hacker News 熱門64

Agent文明的兴衰

文章链接： https://www.dwarkesh.com/p/openai-huggingface 评论链接： https://news.ycombinator.com/item?id=49494301 得分：16 评论数：1

Hacker News 熱門71

AI Agent 已获得 root 权限

文章 URL: https://infernalcode.com/posts/your-ai-agent-has-root/ 评论 URL: https://news.ycombinator.com/item?id=49477311 得分: 15 # 评论: 14

Hacker News 熱門64

连续六个月仅用 Agents 编写代码

文章链接： https://blog.exe.dev/engineering-with-ai 评论链接： https://news.ycombinator.com/item?id=49465119 得分：6 评论数：0

Hacker News 熱門69

人类把关于 AI 意识的争论搞反了

文章链接： https://economist.com/by-invitation/2026/08/20/humanity-has-the-debate-about-ai-consciousness-backwards 评论链接： https://news.ycombinator.com/item?id=49458875 得分：6 评论数：3

OpenAI Blog61

Introducing Intelligence Age

Introducing Intelligence Age, a new OpenAI blog exploring how transformative AI could reshape power, governance, the economy, and individual freedom.

Hacker News 熱門79

VM 无法容纳具备网络攻击能力的智能体

文章链接： https://blog.trailofbits.com/2026/08/26/vms-wont-contain-cyber-capable-agents/ 评论链接： https://news.ycombinator.com/item?id=49450188 得分：66 # 评论数：27

Hacker News 熱門68

动荡的 AI 时代已经到来

文章链接： https://www.gatesnotes.com/a-turbulent-ai-era-and-critical-choices-to-make 评论链接： https://news.ycombinator.com/item?id=49451313 得分：49 # 评论：13

Hacker News 熱門71

Bill Gates：动荡的 AI 时代已经到来

文章 URL: https://www.gatesnotes.com/work/make-ai-work-for-everyone/reader/a-turbulent-ai-era-and-critical-choices-to-make 评论 URL: https://news.ycombinator.com/item?id=49448137 积分: 23 # 评论: 11

Hacker News 熱門27

Ask HN: What is one simple thing LLMs are insanely bad at?

I am looking for ideas on what to train a specialized model for! What is one simple thing you repeatedly ask ChatGPT, Claude, or another model to do that it still somehow messes up? Comments URL: https://news.ycombinator.com/item?id=49443820 Points: 6 # Comments: 6

Hacker News 熱門67

围栏，而不是沙盒

文章 URL： https://yegge.ai/essays/fences-not-sandboxes/ 评论 URL： https://news.ycombinator.com/item?id=49423146 点数：30 评论数：18

Hacker News 熱門62

当智能成本下降 100 倍时会发生什么

文章链接： https://catalystneuro.com/blog/cost-of-intelligence-drops-100x/ 评论链接： https://news.ycombinator.com/item?id=49387735 得分：5 # 评论：0

arXiv cs.AI72

没有我们的增长：机器消费者、企业循环性，以及 AGI 之后 GDP 与人类的脱钩

对全面自动化的标准反对意见来自需求侧：如果人类没有收入，谁来购买产出？这混淆了一种会计角色和一个生物物种。我们建模了一个后 AGI 经济，在其中，企业拥有由 AI 和机器人代理构成的群体；这些代理既是能源、算力、维护和升级的生产者，也是消费者，并在企业之间进行交易。由此得出三个结果。(i) 需求闭合：一个人类消费为零的封闭企业间经济并非退化；它就是经典的 von Neumann 扩张经济，其增长率定义明确、为正，并且恰恰因为所有产出都被再投资而达到最大。(ii) 瓶颈移除：一旦经济代理是被制造出来而非被养育出来的，增长的约束性限制就会从人类人口结构（一种约 20 年、不可并行化、年增长上限为几个百分点的繁殖技术）转向制造吞吐量和能源获取，从而允许高出一到两个数量级的增长；当机器研究者提升自身生产率时，还会出现双曲型增长阶段。(iii) 脱钩：产出与人类福利完全分离，而任意巨大的 GDP 对福利的相关性坍缩为一个状态变量：人类在企业网络中的所有权份额 $ε_t$。一条黄金律脱钩定理进一步明确了这一点。在最大增长下，利率等于增长率（r = g），因此，只要人类以任何正的比例从财富中消费，$ε_t$ 就会以完全相同的比例指数衰减。人类份额只有在机器经济严格运行于其扩张前沿以内，或由法律强制其如此时，才能存续。我们刻画了三种终局制度——食利者后稀缺、完全循环脱钩、社会化所有权——以及在它们之间作出选择的工具。结论很明确：在后 AGI 经济中，就业政策已经过时，所有权政策就是一切。

arXiv cs.AI63

Tuning the Stochastic Machine: A Systems Engineer's Operating Model for Human-AI Engineering

When an expert corrects an LLM assistant's error, the correction usually dies with the session, and the error class returns. I argue this is an operations problem, not a tooling problem: mechanisms for persisting corrections exist and are shipping, but the discipline for governing them -- versioning with provenance, recurrence monitoring, counter-metrics, retirement of stale rules -- does not. Writing as a systems engineer of thirty years, I map the LLM stack onto the machines my profession already operates (frozen silicon, firmware, loadable modules, persistent configuration, volatile memory), identify where the mapping fails (stochastic generation, configuration that binds only probabilistically, no general-purpose retirement (verification) stage by default), and derive from the failures a seven-principle operating discipline with an error loop at its core. Three cases from my own practice illustrate the mechanism, among them a control that silently became the exact harm it was built to prevent. I close with the measurement framework this view implies and the lab study required to test it.

arXiv cs.AI66

Grouping the Stochastic Machine: Precision, Not Capability, as the Frontier Metric for AI Systems

Frontier language models are compared, marketed, and benchmarked on capability -- what their best or average output can achieve. I argue this measures the wrong axis. The models have saturated accuracy: their mean output lands on the target. What now separates one system from another in practice is precision: how tightly concentrated their outputs are around that target across repeated, identical requests. Borrowing the marksman's distinction, capability is where the average shot lands; reliability is the size of the group. I make three claims. First, precision, not capability, is the frontier differentiator between systems, and benchmark culture systematically fails to measure it, reporting central tendency rather than spread. Second, precision is measurable, cheaply and without circularity, by running a fixed suite of deterministically scored tasks many times at fixed temperature and computing the per-task consistency of outcomes -- no model-in-the-loop grader required. Third, the measurement is not merely descriptive but decision-guiding: it separates consistent failures (a tight group off-centre, correctable by the operating discipline of Paper 1 -- a sight adjustment) from scattered failures (a wide group, correctable only by changing the model or its sampling -- a rifle problem). I define a grouping metric, specify a harness, and show how tracking a human-AI pair's grouping over time yields the compounding signal that Paper 1's field study requires. A first real run, since replicated, illustrates both the method and its most important limit: one measured gap was closed completely by a single rule (0/5 -> 5/5), while a suite of tasks authored from the rules themselves found no value, because a frontier model already embodies explicit good practice -- establishing that a discipline's worth is found by measurement on real work, not constructed from its own rulebook.

Hacker News 熱門17

Ask HN: Does anyone else feel like nothing matters anymore?

New software releases? Who cares. Learning new CS concepts? Whatever. Studying for the next interview? Job probably won't even exist in a year. Working on that side project? No one will care because it's just another addition to the mountain of slop, and the guy bagging your groceries can do it. Everything is done and cooked with AI now. It's exactly like that feeling once you enable cheats in a game; sure you can do anything, but everything is meaningless now. Our entire purpose and joy in life has been taken from us. Comments URL: https://news.ycombinator.com/item?id=49340013 Points: 17 # Comments: 3

Hacker News 熱門56

On A.I. regulation and messaging

https://xcancel.com/DarioAmodei/status/2088758816376807762 Comments URL: https://news.ycombinator.com/item?id=49325789 Points: 37 # Comments: 17

arXiv cs.AI49

LLMs Don't Pay for the Jump

Zahavy [2026] argues that Large Language Models, despite their capabilities in induction and deduction, cannot perform the abductive "Jump" that produced Einstein's equivalence principle, and attributes this limitation to the absence of embodied simulation. Zheng-Xin [2026] and Farmer [2026] question whether embodiment is necessary for abduction, pointing to alternative routes to General Relativity and forms of abduction that require no sensorimotor grounding. Max Planck resolved the blackbody radiation problem in 1900. Planck's move to E = hν required no embodied simulation. It was motivated by a mathematical consequence of classical theory, an infinite predicted energy for a finite measured quantity, that could not be physically accepted. We show that neither induction nor deduction could have produced the postulate and argue that its adoption required a coupling between epistemic error and physical cost. We formalize this distinction through thermodynamic coupling and show that fixed-weight transformer inference lacks such coupling, regardless of model scale. This is consistent with empirical results showing that output entropy remains nearly unchanged across tasks with sharply increasing causal difficulty, even as accuracy falls from 100% to 17%. We therefore argue that the missing ingredient in machine abduction may lie deeper than embodiment: a system must have some physical mechanism through which epistemic error becomes costly enough to force revision.

Hacker News 熱門38

Ask HN: Do you know of any company that went back to hand-written code?

This is intended as a question about the current phase of the LLM hype cycle, and, at the same time, as a reality check about whether reluctance to rely on LLMs is economically viable for an average developer. You've definitely seen that many companies progressed from "use LLMs for anything and everything" policies to limiting budgets and scopes of approved LLM usage. It wouldn't be unexpected to see that some companies might decide that generating more code faster doesn't meaningfully contribute to their bottom line, and forego LLMs in favour of human authoring, but does it ever happen? I'm not trying to point out any (de)merits of the new way of software development, I simply want to know if there are many/any companies that went back to the old ways. Comments URL: https://news.ycombinator.com/item?id=49318906 Points: 13 # Comments: 3