Against "solving alignment" as the major frame for AI safety and security