Anthropic: Claude-led AI R&D tasks jumped from 1% to 26% in six months

Hesamation · x · 2026-09-18

Anthropic published three internal metrics to track AI's role in building the next version of itself: share of AI R&D done by AI, quality of agent oversight, and compute allocation.

Key numbers:

Anthropic argues any frontier lab could publish the same measures for third-party verification, to narrow the gap between what frontier labs know and what the public knows.

Related event: Anthropic Discloses Internal Metrics: Claude Now Drives 26% of Its R&D(41 posts)→

Original post →

More from AGI Musings

AGI Musings channel →