METR finds Anthropic may have crossed AI acceleration warning threshold in July

Nonprofit METR has found that Anthropic may have crossed its own AI acceleration warning threshold on July 15, 2026, after roughly eleven weeks of access to internal research data. METR says the rate at which Anthropic's models are improving has roughly doubled, and about 80 percent of that acceleration traces back to AI being used in the company's own research. Anthropic redacted much of the evidence METR rated as most informative.

Audit access and findings

From August 12 to October 29, 2026, a team of 2.5 full-time METR staffers received access to benchmark results from unreleased models, training metrics, employee surveys and internal analyses, plus roughly 15 interviews with Anthropic researchers. METR took no money from Anthropic but worked on Anthropic devices for security reasons.

Since October 2025, Anthropic's model test scores have been climbing about 1.9 times faster than they did between March 2024 and October 2025, METR says. The measurement is based on the AECI, a composite benchmark metric. Before October 2025, Anthropic's models gained about 15 AECI points per year; since then, the rate has hit 29.

AI in research and uncertainty

METR is confident the acceleration is not driven by faster growth in training compute. Instead, the auditors estimate that about 80 percent of the current speedup comes from AI being used in Anthropic's own research. Employees said in informal Slack polls from October 2026 that the latest internal model made them roughly twelve times more productive, but METR treats these self-assessments as "very loose evidence" that is unrepresentative and likely to overstate gains.

AI is not doing fully autonomous research yet, according to METR. The acceleration comes mainly from AI acting as an assistant, but there are "signs of life" that independent contributions could play a bigger role soon. METR found no bottlenecks that would slow the pace over the next twelve months, which it says means there is a "significant risk" that the acceleration will keep growing over the coming year.

METR stresses the results are imprecise. The true acceleration factor could be as low as 1.3 or as high as 3. Factoring in OpenAI, Google and xAI models, the industry-wide number drops to 1.4, possibly because Anthropic lagged behind competitors in 2025 and a catch-up effect is skewing the figures.

Redactions and warning threshold

Anthropic had unilateral redaction rights over all non-public information and used them heavily. According to METR, the majority of the evidence rated as most informative was stripped out, including forecasts for future models, breakdowns by training steps and the bottleneck analysis. Anthropic says the redactions were "aimed at removing confidential information that could accelerate other AI developers." METR writes that the public has to rely on its judgment "much more than would be ideal."

In a companion review, METR examines Anthropic's risk report from August 2026. In that report, Anthropic publicly stated the acceleration was "less than a factor of 2." The internal, unredacted version implies a value of 1.97, though, and Anthropic itself calls it "plausible" that the threshold has been crossed. METR considers it possible that the warning level was already reached on July 15 and accuses the public report of understating both the acceleration and the uncertainty around it. METR currently shares Anthropic's assessment that the acute catastrophic risk is "low," but it cannot rule out another sharp acceleration over the next six months.

Sources: The Decoder