Key points
- METR analysis indicates Anthropic is improving AI capabilities at roughly double the rate of the previous year.
- CEO Dario Amodei published an essay on 12 September urging the industry to slow the pace of capability improvements.
- The research group reviewed internal data and interviewed staff before Anthropic’s planned Wall Street debut.
Model Evaluation and Threat Research reported that Anthropic is developing artificial intelligence systems approximately twice as quickly as it was one year ago.reported This finding emerged from an analysis released on a Thursday by the Berkeley-based nonprofit, highlighting a contrast with public statements from Anthropic leadership regarding development speed.An independent group found Anthropic is building AI twice as fast as last year, even though its boss says the industry should slow down.
Evidence: reported
Anthropic’s AI development has doubled in speed despite slowdown calls, research group saysaifleet-fastcompany
The claim
METR estimated that Anthropic is improving the capabilities of its models to perform well specified tasks around twice as fast as the previous year. The researchers believed that much of this accelerated development stems from the use of AI itself to assist in the improvement process. This assessment occurred while Anthropic prepared for its initial public offering on Wall Street.
The response
Anthropic CEO Dario Amodei called for the AI industry to slow its work so that safety measures could catch up with technological progress. In an essay published on 12 September, he urged the global AI community to reduce the pace at which they improve the capabilities of AI models.reported Anthropic did not immediately comment on the report in response to inquiries.
Evidence: reported
Anthropic’s AI development has doubled in speed despite slowdown calls, research group saysaifleet-fastcompany
The bottom line for teams
The findings are significant because details about AI technology and its associated risks have been provided primarily by the companies themselves. These firms have been encouraged by the administration of President Donald Trump to police themselves on matters of safety. The independent review offers an external perspective during a period of heightened debate over potential catastrophic consequences.
What the paper shows
The research compares the current rate of capability improvement in Anthropic’s AI systems to the rate from a year ago. The reported result is that development is around twice as quick.reported A stated limitation is that the findings rely on internal data reviewed by METR researchers, who also interviewed staff. The analysis suggests AI tools are driving their own improvement cycle, but does not quantify the exact safety implications of this speed.
Evidence: reported
Anthropic’s AI development has doubled in speed despite slowdown calls, research group saysaifleet-fastcompany
Why it matters
- Evaluators can benchmark internal development cycles against an independent external assessment of a major vendor.
- The source notes that details about technology and risks have been provided primarily by AI companies.
- The source states that companies have been encouraged by the Trump administration to police themselves on safety.
- The source reports that METR believed AI itself helped cause the faster development speed.
Topics: Anthropic, METR, AI safety, model evaluation, IPO
Updated
Then / now / next
AI companies primarily provided details about their technology and risks without independent verification.
An independent group has assessed internal data and found development speeds have doubled.
Anthropic plans to go public later this month, potentially bringing more scrutiny to its development practices.
Story timeline
- Then AI companies primarily provided details about their technology and risks without independent verification.
- 05 Nov 2026 Anthropic’s AI development has doubled in speed despite slowdown calls, research group says
- Now An independent group has assessed internal data and found development speeds have doubled.
- Next Anthropic plans to go public later this month, potentially bringing more scrutiny to its development practices.
The strongest doubt
Counterpoint from the sources
Anthropic did not provide an immediate comment on the report, leaving the company's official stance on the findings unclear.
Ask the article
Local keyword search of this article; no generated answers.Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Questions readers ask
Did Anthropic confirm the METR findings?
Anthropic did not immediately comment on the report in response to inquiries.
What caused the doubling of development speed?
METR researchers believed much of the reason was the help of AI itself in improving models.
When did the CEO call for slowing down?
Dario Amodei published an essay on 12 September urging the industry to slow the pace of improvement.
Connected stories
Reading receipt
- METR analysis indicates Anthropic is improving AI capabilities at roughly double the rate of the previous year.
- CEO Dario Amodei published an essay on 12 September urging the industry to slow the pace of capability improvements.
- The research group reviewed internal data and interviewed staff before Anthropic’s planned Wall Street debut.
Still open: How exactly is AI assisting in the acceleration of model development?
Still open: What specific safety measures are lagging behind the increased development pace?