Anthropic CEO calls for pacing frontier AI development, warns of agent swarms

Anthropic CEO Dario Amodei has outlined a plan to slow the development of frontier AI models, describing the goal as “pacing the frontier,” according to TechCrunch.

The same remarks were covered the same day by TechRadar and The Washington Post, which reported that the Anthropic chief is calling on the AI industry to slow down. TechRadar highlights the timeline he attached to the warning: within six to 12 months, a swarm of AI agents could be capable of taking over the entire internet.

That timeline is what makes the warning notable. Safety debates have often placed the risk in a distant future; six to 12 months turns an abstract existential concern into an engineering problem measured in quarters.

TechCrunch also notes that OpenAI’s Sam Altman appears to agree it is time to pace the frontier, even though the two labs compete directly at the top of the model market. That shared language is itself worth watching.

The report then asks the practical question: what would pacing actually look like? Slower training runs, a longer gap between releases or more evaluation before deployment are all open possibilities, and the coverage leaves the question unanswered.

For the industry, the signal matters more than the rhetoric. When the head of a leading frontier lab publicly talks about slowing down, the competitive story can shift from who builds the strongest model to who can keep capability contained.

What to watch next is whether the call hardens into concrete mechanisms — evaluation thresholds, release cadence or coordination between labs — and whether rivals adopt the same language.