Anthropic Publishes First Transparency Metrics: Claude Leads ~26% of Its Internal AI R&D
TheZachMueller · x · 2026-09-18
Anthropic unveiled three transparency metrics for tracking AI self-development: how much AI R&D is done by AI, how well AI agents are overseen, and how compute is allocated, with an internal snapshot for each.
The headline figure: Claude "leads" about 26% of Anthropic's AI R&D work. Nathan Lambert noted the report never defines what counts as "AI R&D work," making the number hard to interpret, but called it a step in the right transparency direction. Anthropic says any frontier lab could publish the same measures, with third-party verification.
More from Models
- Community unmasks stealth model Union Alpha as Unbiased's Pareto 26.9 in one day — gaganghotra_ · 2026-09-18
- Stealth model Union Alpha unmasked in a day as eval outfit's Pareto 26.9 — karminski3 · 2026-09-18
- Emad Mostaque hails an 'Opus 4.5-level' model that runs on 8GB RAM — ccerrato147 · 2026-09-18
- Epoch AI audits 15 AI benchmarks, finds 9 flawed — 46% of sampled HLE questions broken — morqon · 2026-09-18
- Astra computer use formats Google Docs autonomously, leaving users impressed — var_epsilon · 2026-09-18
- GPT-6 Astra reportedly tops FrontierSWE at 65.5%, nearly doubling predecessor — charliermarsh · 2026-09-18