ANTHROPIC
AI Accelerating AI Development: Alarming METR Report on Anthropic
An 11-week audit by research organization METR reveals that Anthropic's AI models are boosting R&D velocity, improving twice as fast as last year.
A newly published report by AI safety organization METR, following an 11-week deep dive inside Anthropic, reveals stark evidence of artificial intelligence accelerating its own technical development. The findings show that Anthropic’s models are streamlining research and engineering tasks, causing capability milestones to arrive roughly twice as quickly as the previous year.
While Anthropic publicly advocates for cautious pacing in AI development, its internal systems demonstrate powerful recursive self-improvement capabilities. This dynamic highlights mounting challenges for regulators and safety researchers tracking runaway technological loops.