AI Accelerates Its Own Development, Study Finds
An independent evaluation by METR says Anthropic's models are improving nearly twice as fast as before, with AI driving most of the speedup.
Updated on Nov. 5, 2026 in Artificial Intelligence
Live Poll
Do you think AI should be strictly regulated now, before it advances further?
METR found that Anthropic's AI models have improved 1.9 times faster since October 2025 than previously. The nonprofit estimated AI accounts for perhaps 80% of the current acceleration.
Why it matters
The findings suggest AI is playing a growing role in building more powerful versions of itself, which could shape the pace of progress across the industry. METR's assessment found no convincing evidence of major bottlenecks that could slow that pace over the next year.
METR's 11-week assessment measured a 1.9-times speedup in Anthropic model capability gains since October 2025, with AI credited for roughly 80% of the acceleration. METR ruled out increased training compute scaling as the cause, but it is not clear how long the faster pace will persist.
The players
METR
METR is an independent research nonprofit that evaluates the capabilities and risks of frontier AI systems, and it produced this assessment.
Anthropic
Anthropic is an artificial intelligence company that develops the Claude family of AI models, whose models were the subject of the findings.
The details
METR's assessment drew on confidential research, interviews, and performance data from unreleased models, and it ruled out an acceleration in the scaling of training compute as the cause. The group concluded that AI is not yet capable of making highly significant, fully autonomous contributions to the most important stages of model training.
Timeline
October 2025: Anthropic model improvement begins accelerating.
November 5, 2026: METR publishes its findings.
The Tech Race
METR's evaluations of frontier systems have become a key independent check on how fast labs such as Anthropic are advancing, and this finding extends that tracking to show AI increasingly driving its own progress. The result positions Anthropic against rival labs in an arms race where the pace of self-improvement is becoming a central competitive variable.
Faster model improvement could bring new AI capabilities to users sooner, affecting the tools people rely on for work and daily tasks. Users may also see shifts in pricing and access as labs release more capable systems on shorter cycles.
The takeaway
Faster AI progress means the tools available to you may change quickly, so it is worth checking how new releases affect the features and costs of the services you use. Staying informed about independent evaluations like METR's can help you judge the pace of change.
Further reading
Explore more on Artificial Intelligence developments and how AI tools are advancing across the industry.
Live Poll
Do you think AI should be strictly regulated now, before it advances further?
