Sam Altman warns AI progress risks losing control to AI or concentrating power in one lab, company or country

The author identifies two catastrophic risks from AI progress: losing control of the future to AI, and excessive concentration of power in one person, company, lab, or country. The author calls for alignment and safety techniques to stay ahead of capabilities, welcomes a federal framework with consistent frontier AI safety requirements, and supports independent auditors—without waiting for legislation. OpenAI now writes explicit safety cases before frontier reinforcement learning runs, beyond pre-release safety work. The author urges industry-wide shared standards for misalignment and monitoring, says "pacing" means slowing, not stopping, progress, and asks government help for international coordination.

✨ AI summary of the linked source — not editorial content.

Explore the topic

• Risks
• AI