Sam Altman: regulate AI like nuclear weapons

What is at stake

When the creator of a new technology calls for strict regulation of it, the world should listen carefully. That is exactly what Sam Altman, the head of OpenAI, is doing. At the Stripe annual conference in San Francisco, he compared artificial intelligence to nuclear weapons and demanded a global oversight body modelled on the IAEA. The man who triggered the AI boom with ChatGPT is saying: without rules, the technology will spiral out of control.

The comparison Altman draws

The IAEA was founded in Vienna in 1957 and has more than 150 member states. It sends inspectors to countries using nuclear material and ensures that the knowledge is not diverted to military use. Since 1945, no nuclear bomb has been used as a weapon. Altman says: “Hardly anyone would have thought that possible at the time.” He believes a global AI authority could achieve something comparable – and he considers the need urgent.

The contradictions of the warner

The contradiction is plain to see. Altman himself leads the very company that is currently driving the fastest AI development in the world. OpenAI is the leader in language models, has close partner Microsoft on board with some 13 billion dollars invested, and is now integrating AI into nearly every Microsoft product. Calling for regulation while setting the pace of an entire industry invites suspicion – or it signals that he knows how serious the situation truly is.

What this means for Europe

Europe is working on the AI Act, the G7 have set up an AI working group, China has its own rules, the United States is debating voluntary commitments. What is missing is precisely the global institution that Altman is calling for. An IAEA for AI would need cross-border access to training runs, data centres and model weights. That is politically hard to imagine. But the alternative, by Altman’s logic, is not no regulation – it is a development that no one can control any more.

I wrote the full piece for Handelsblatt.