Anthropic says Claude leads 26% of internal AI research work
TechnologyBreaking
Archived — This article has been archived. The information may be outdated.

Anthropic says Claude leads 26% of internal AI research work

Livemint··18 Sept

Anthropic said on September 17, 2026 that Claude leads 26% of AI research and development on an Epoch AI scale, up from 1% in March. AI collaborated with humans on more than 90% of research work as of August. About 30,000 AI agents worked on its platform in August, with about one in 47,000 actions blocked after screening.

Prism

What It Means For You

  • Customers using Claude may see faster internal tool updates if AI agents keep gaining share.
  • Safety teams get a larger compute slice when AI itself runs safety research.
  • Regulators tracking autonomous coding will watch Anthropic's blocked-action rate.

What's Happening

  • Anthropic published the metrics in a Thursday blog post.
  • OpenAI disclosed six misalignment cases the prior day.
  • Claude is not fully autonomous in any measured workflow, the company said.

Epoch AI scale for Claude share

  • Epoch AI is an independent nonprofit tracking frontier AI development.
  • More than a billion agent decisions were screened in August.
  • Researchers warn autonomous agents may diverge from creator intent.
all-newstop-storiestrending-now

More in Technology