Sam Altman Pitches IAEA-Style, U.S.–Led AI Safety Forum After G7 Talks

Calls for supranational tech governance echo the 1957 founding of the IAEA—born when Eisenhower’s 1953 “Atoms for Peace” speech crystallized concern that a single state could not contain nuclear risk. Similar déjà-vu surrounded the 1865 International Telegraph Union and the 1944 Chicago Convention on aviation: each arose when infrastructure scaled faster than national rules. Altman’s push reflects a longer trend—U.S. firms seeking to write first-mover rules that lock in both market share and safety narratives before Chinese or EU models mature. Unlike uranium centrifuges, however, AI weights are duplicable and inspectable only via voluntary access, raising enforceability doubts. Whether this bid succeeds may dictate if AI governance in 2126 resembles the globally harmonized nuclear regime or the fragmented, multipolar internet that followed the 1998 U.S. export-control split. In that century-long arc, today’s skirmish over who drafts the rulebook could matter more than any single model release.