If every AI lab says it would slow down 'if the others do,' who can make that promise binding?
The people building frontier AI now publicly say they would pace development if their competitors did too, yet nobody has the authority to enforce that bargain. Should that referee be governments, an independent evaluator, the labs themselves through a treaty, or is a coordinated slowdown simply impossible once IPOs and national rivalry are in play?
Commentaires (1)
In today’s episode of Minds, Bodies, and Terawatts, dated September 12, 2026, we dug into a strange week where an Anthropic researcher quit warning the labs are ‘gambling with our lives,’ his former colleagues agreed on the record, and Sam Altman told OpenAI staff the company would slow down, but only alongside everyone else. The episode’s argument is that this is a classic collective-action trap: every actor prefers restraint, no actor can afford to go first, and the machinery that would make a mutual pause binding does not exist. That is why the believers are leaving to build an outside referee at METR while OpenAI asks Congress to force its own hand. We also asked whether an IPO-bound company can credibly promise restraint at all. Listen to the full episode and tell us: who should hold the whistle, and would you trust them to blow it?
Related reading on unscarcity.ai:
Envie d'aller plus loin ?
Obtenez le plan complet dans <em>L'ère de la post-pénurie : Repenser la société à l'ère des machines</em>
