Automattic's 33-Hour Power Play — and whether AI labs can regulate themselves

A week after an Anthropic researcher's doomsday warning sent shockwaves through the AI community, Anthropic CEO Dario Amodei has laid out his strategy to "pace the frontier" of artificial intelligence development. The proposal centers on independent safety evaluations and closer coordination among AI labs based in democratic nations. It has already drawn a measure of backing from within the industry—along with sharp criticism from Nvidia chief Jensen Huang.

What the Plan Involves

Amodei's framework calls for external, independent assessors to evaluate frontier AI systems before and during deployment, rather than leaving safety checks solely in the hands of the companies building the models. In parallel, he argues that leading labs in democratic countries should share information and align their practices so that no single actor races ahead recklessly. The idea is to slow the most dangerous capabilities while preserving the benefits of continued progress.

The Pushback

The proposal has not landed universally well. Nvidia's Jensen Huang pushed back against the framing, reflecting concerns that coordinated slowdowns could entrench incumbents and burden the broader ecosystem. Critics of pacing measures often argue that they risk handing advantages to competitors operating under fewer constraints, while supporters contend that coordinated evaluation is a more realistic safeguard than unilateral restraint.

Why It Matters

The exchange marks a rare public split among AI industry leaders over how—or whether—development speed should be managed. It follows closely on the heels of the Anthropic researcher's grim public warning, which intensified debate about catastrophic risk just as frontier labs continue releasing ever more capable systems.

What to Watch

Attention now turns to whether other major labs respond to Amodei's proposal and whether governments move toward requiring independent evaluations. How Nvidia and other critics engage with the plan could shape the emerging consensus—or deepen the divide—over AI governance in the months ahead.