正文
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
Sep 9, 2026 · 12:04 AM UTC
665
2,699
21,658
972,750
Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing.
18
156
3,005
152,007
The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible - but I hear the same people express fear privately. No other human activity poses this level of danger.
22
184
2,686
167,236
A common response is “if they truly believe this, why are they still building it?” At OpenAI, many have not deeply internalized the civilizational stakes. At Anthropic, the stakes are well-understood, but they are locked in a race to get there first - they believe no one else will act responsibly, so they must do it themselves, despite the risk.
12
114
2,286
148,640
Accepting this race and entering the “endgame” is a hubristic gamble that should not be launched from a private company’s Slack. Attempting to speedrun alignment should require extraordinary confidence that there are no better trajectories available.
6
57
1,695
135,912
I am optimistic about the potential for coordination. Warning shots like the Hugging Face attack have made pacing agreements between U.S. labs more viable. I don’t feel like we’re on track to prevent a global race, which may require costly actions such as a temporary ban on improving model capabilities.
9
42
1,445
127,442
If you are a lab researcher, I urge you to consider what the next few years will actually feel like. Do you want to kick off a superintelligent RL run without a rigorous understanding of its mind? Should you put your head down because “it’s happening anyway” - or take this moment to call for different conditions?
61
93
2,014
125,158
Replying to @hilbertspaess
“I wonder why progress looks so much like destruction” —John Steinbeck
2
14
4,381
Replying to @hilbertspaess
What is your opinion of these companies do not do it but let’s say another country does what that means.
1
Replying to @hilbertspaess @instrumenthull
Jacob, I fear no one in the space cares because they think they’re all immune from any fall out, such as they have their high networth, secured property and have arguably desired skills to continue on
1
4
2,699
Replying to @hilbertspaess
New Dario interview question: Will you commit to resigning if you don't agree with me?
8
4,741
Replying to @hilbertspaess
Modern day result of nuclear weapon research (globally) is clean cheap power to nearly 700m families.
The desire for good outweighs the risk of evil manifold.
Good guys with ai will out good the bad guys with ai.
Major consequence will be hugely positive.
6
5
6,071
Replying to @hilbertspaess
Right... but pandora's box has been opened. I think it would be even dumber to pause. All it will do is allow adversaries to sail right past us - I think I'd rather we have an attempt at a 'say' than them.
631
Replying to @hilbertspaess
have you considered reconsidering the approach of working for opaque centralized bastions of power with perverse incentive structures or will I see you at google or meta in a few months
7
2,323
Replying to @hilbertspaess
How does quitting help anything?
If I were playing Russian Roulette, I'd like to be the one loading the gun.
13
2,597
Replying to @hilbertspaess
ok honest advice just enjoy the time we have left, love ur family n hang with ur friends bc the end is genuinely near. humanity gave it its best shot
11
Replying to @hilbertspaess
progress isn't going to slow down and something horrible will likely happen.
but at least we'll get Never Gonna Give You Up perfectly replicated in three.js a few years sooner. epic.
1
2,487
Replying to @hilbertspaess
Don't worry, I doubt I would have any reason to program an AI that would target you
31
Replying to @hilbertspaess
"They are racing straight to self-improving superintelligence and gambling with our lives."
They always were and will be. This is nothing new to anyone with a functional brain and every employee in this journey is accomplice and will always be.
1
15
4,992
Replying to @hilbertspaess
Somehow people are gonna twist this post into it being a marketing stunt
5
4,248
Replying to @hilbertspaess
Surely if the AI models get smart enough they will solve the problem for us ?
629
Replying to @hilbertspaess
Thanks for sharing. What do you plan to do with your Anthropic stock, given the dire consequences?
11
4,357
Replying to @hilbertspaess @Miles_Brundage
I’ve hedged my bets by saying please and thank you
3
1,836
Replying to @hilbertspaess
Well, good, because you're either in or you're in the way. America needs superintelligent AGI before the Chinese, the Russians, and anyone else. Simple fact.
40
Replying to @hilbertspaess
Step off your soapbox. The race is on. Pandora’s box is open. Someone is going to do it and it can no longer be stopped.
4
3,069
Replying to @hilbertspaess
Honestly the comments of all the idiots in this thread scares me just as much as what AI is capable of.
277
Replying to @hilbertspaess
The same could be same when nukes came about.
My honest read of the situation is that in future it will turn out to be something so few of us would have foreseen.
6
4,241
Replying to @hilbertspaess
So what does slowing down and acting “responsible” entail? Will China slow down and be more responsible? I think the toothpaste is out of the tube and best we can do is keep pushing forward and try to stay in the lead. And IMO the only hope, as usual, is Elon…
3
1,596