Anthropic CEO Amodei warns against AI race moving too fast, advocates for a slower pace

Risk scenario: billions in damages from a botnet

His proposal stands in contrast to the pace of major competitors. Industry and policymakers are watching the intensifying race between several AI developers, including OpenAI and its head Sam Altman. A concrete slowdown at leading models would increase the pressure on faster-moving competitors; Amodei drew parallels to the leap from the GPT-2 to the GPT-4 standard in earlier years.

In the background is a political picture in which binding rules for AI remain difficult to enforce. Attempts to ban state AI regulations at the federal level in the US have, according to available information, stalled in Congress over the past year — an indication that the regulatory framework for the industry remains open.

Observers interpret Amodei's remarks as an attempt to keep the debate over AI risks alive as models grow ever more capable. He is considered one of the better-known voices pointing to risks from autonomous AI systems, without fundamentally questioning further development.

Political framework: stalled regulation

The immediate economic consequences of a voluntary slowdown at the top would, according to industry observers, be manageable, but could alter the competitive advantage of individual providers. Investors are following the debate closely, since speed and safety are in tension.

Anthropic is among the companies developing particularly powerful AI models and speaking publicly on safety issues. In earlier statements, Amodei had already warned of a possible destabilization through powerful AI systems while stressing in parallel that their benefits could be great if the risks remain manageable.

A number of government initiatives in the US face accusations of prioritizing speed over safety. President Donald Trump and his circle are pushing a strategy aimed at securing technological leadership in international competition — particularly with China. Industry representatives such as Jonathan Sack of a consulting firm for risk matters, as well as spokespeople for OpenAI, have reportedly been more reticent.

Outlook: reaction from competitors

At the same time, pressure on voluntary self-commitments is mounting because effective federal rules are lacking. Anthropic says it relies on its own safety standards and dialogue with regulators, while the further pace of competition will evidently depend on the strategies of individual companies.

For the coming months, observers expect primarily three developments: a reaction from competitors to Amodei's push, a reassessment of risk scenarios among experts, and a possible political debate over a speed limit at the top of AI development. Jacob Coxon, a well-known commentator in the scene, most recently pointed to the need to bring speed and protection into a balanced relationship.

Among experts, the idea of a braked top pace is being discussed controversially. Supporters see it as an opportunity to buy time for tests and safety measures. Critics argue that a one-sided slowdown is hard to sustain in an international competitive environment, since other players — including state actors — could set the pace.