Ex-OpenAI Employee Proposes Global AI Pause Until 2040 in Optimistic Follow-Up to Extinction Warning

Ex-OpenAI employee Daniel Kokotajlo, who previously warned of AI-driven extinction, now proposes a hopeful global plan to pause development of superhuman AI until 2040, paired with radical transparency measures — though he admits major resistance from AI companies is expected.

• Former OpenAI employee Daniel Kokotajlo, whose 'AI 2027' scenario warned of human extinction or authoritarian AI control, is now releasing a follow-up vision called 'AI 2040: Plan A' that offers a more optimistic path forward.
• The new plan calls for an international agreement to delay the development of AI smarter than any human until 2040, along with radical transparency measures including public access to advanced AI models currently kept locked away by companies.
• Kokotajlo acknowledges the plan faces major hurdles, predicting that AI companies will resist the proposed restrictions, but hopes terms like 'Plan A' will shape global policy debates before a crisis forces world leaders to act.