Altman backs coordinated AI slowdown with rivals

The shift this week was not a new disaster but a change in who is talking about AI's most extreme risks. For decades, warnings about AI posing a grave threat to the species stayed at the fringe, voiced occasionally by figures like Elon Musk, Sam Altman, or Geoffrey Hinton and rarely by the public at large. What changed, per the article, was a drumbeat of resignations from employees at Anthropic, OpenAI, and Google DeepMind, combined with coverage of the Hugging Face incident and firsthand experiences with AI agents, which seems to have widened the Overton window on loss-of-control concerns. The timing mattered too: Anthropic is on the verge of an IPO and OpenAI is edging closer to one.

Once the debate moved, the responses split sharply. Altman and Amodei converged on the idea of a coordinated slowdown, with Amodei floating an antitrust exemption and independent on-site evaluators, and Altman noting that pacing does not mean stopping. Lawmakers followed with proposals ranging from an outright ban on artificial superintelligence to a duty to prevent catastrophic harms, and former President Barack Obama urged Democrats to center AI governance. The pushback was just as direct. Trump criticized Amodei by name, Speaker Mike Johnson called the fear media-driven, and Chinese state media labeled Amodei's proposals self-serving Cold War tactics. The article concludes that prospects for an executive order or actual legislation are poor, at least until after the November midterms.

That leaves the practical questions the article surfaces: whether antitrust concerns are legitimate, whether product liability law is enough, and whether a coordinated safety standard amounts to regulatory capture. Each remains unresolved. Whether a slowdown actually materializes likely hinges on whether voluntary commitments can be given any enforcement mechanism, and on whether the companies involved can agree before the political window shifts.