METR publishes its assessment of whether AI is accelerating AI development at Anthropic

METR, the AI evaluation nonprofit, published a report on a risk assessment in which a team within METR has worked since August with internal Anthropic information to judge whether AI itself is accelerating AI development. It also published a separate review of the AI R&D section of Anthropic's August 2026 risk report.

According to METR, from August 12 to October 29 its staff (about 2.5 full-time equivalents) assessed Anthropic's internal data. Most of the team now believes that, as of October 2026, Anthropic is likely improving its models' capabilities on well-specified tasks at about twice the rate of March 2024 to October 2025, with high uncertainty (it would not be surprised by 1.3x or 3x). It estimates 1.9x for benchmark scores on Anthropic's version of the Epoch Capabilities Index since October 2025 (95% interval 1.5–2.8x), and guesses that about 80% of the current acceleration comes from AI. It says the acceleration was not due to faster growth in training compute, and that it found no clear bottlenecks to further acceleration, though it is not confident in that.

METR says Anthropic had the right to redact non-public information and redacted a majority of the evidence METR found most informative for its conclusions. The post had 864 points on X by November 5.