OpenAI’s Sam Altman calls for an IAEA-style global regime for advanced AI

OpenAI chief executive Sam Altman has proposed establishing an international AI safety forum, explicitly comparing the concept to the IAEA created during the Cold War. Writing in the Financial Times, Altman argued that standard-setting is essential for widespread deployment, warning that looming systems will alter material life more profoundly than electricity.

Under Altman’s proposed blueprint, an international forum staffed by government representatives and technical experts would establish safety baselines and conduct impartial evaluations. In exchange for complying with shared rules, participating nations and their domestic companies would receive guaranteed access to frontier models. The proposal also outlines oversight mechanisms for AI laboratories themselves to counter commercial pressures that fuel reckless capability races.

The push comes as national security intervention becomes tangible rather than theoretical. United States authorities recently instructed Anthropic to suspend foreign access to its Fable 5 and Mythos 5 models, while OpenAI delayed the full release of GPT-5.6 at the request of the government. In his column, Altman framed multilateral certification as a necessary alternative to fragmented export controls and unilateral bans.

The essay formalizes an argument Altman made two weeks earlier during a closed-door dinner at a G7 gathering in France. While Anthropic’s Dario Amodei and Google DeepMind’s Demis Hassabis urged a US-led governance bloc explicitly tailored to isolate China, Altman advocated for a rules-based framework open to any compliant state.

Whether rival superpowers will entrust strategic compute decisions to a collective inspection body, however, remains an untested proposition.

Primary source: Financial Times

Source: habr