Anthropic’s AI Push Speeds Up as Its CEO Urges Restraint
Anthropic’s AI Push Speeds Up as Its CEO Urges Restraint
The tension at Anthropic is no longer merely theoretical: an outside assessment says the company’s models are getting better faster, even as its chief executive publicly argues for restraint at the frontier.
METR, an AI-safety organization that was in contact with Anthropic from mid-August, said the company is “likely improving the capabilities of their models on well-specified tasks at around twice the rate they did between March 2024 and October 2025.” The group attributed that acceleration, with the same qualified language, to Anthropic’s extensive use of AI in research and development.
The chronology matters. METR says it began communicating with Anthropic in mid-August and submitted its final public report on November 3. Between those points, in mid-September, CEO Dario Amodei published his essay, “We Must Pace the Frontier.” The title alone placed Amodei on the side of caution about how quickly the most capable systems should advance.
METR’s finding does not say Anthropic has abandoned that caution. Its assessment is narrower: it concerns progress on “well-specified tasks,” not a blanket judgment on every dimension of AI capability or safety. But it does identify an operational reality that sits awkwardly beside Amodei’s message: AI-assisted R&D may be compressing the time needed to make further gains.
A separate account framed the result even more starkly, reporting that Anthropic was developing AI systems “around twice as quickly as it was a year ago” despite the CEO’s calls to slow the technology’s pace of improvement.
That leaves Anthropic with a difficult public balancing act. The company can argue that caution means governing a faster-moving frontier responsibly, not freezing progress. Critics, however, are likely to see the assessment as evidence that the industry’s safety rhetoric is being tested by the very acceleration its tools enable.
Write a comment