- Who it helps
- Researchers and organizations assessing the strength of external model evaluations.
- Availability
- The October 15 announcement includes an October 16 clarification. This is a limited evaluator arrangement, not general public access to internal systems.
What access can change
Anthropic says METR staff can work alongside its internal risk teams. The arrangement may make evaluation of unreleased capabilities more informative.
Read the publication conditions
The announcement says METR retains freedom to publish, subject to defined redaction categories. Anthropic decides those redactions, while METR can comment on access limitations. This is a material condition when interpreting independence.
What you can do
- Read future evaluation reports together with their access disclosures and publication constraints.
Where the evidence stops
The announcement was checked; METR’s risk-assessment page could not be retrieved. The arrangement’s practical effectiveness remains unverified.
Sources read. Sources checked October 18, 2026. No product workflow or benchmark was executed.
Sources & further reading
Web · official
An update on independent evaluators ↗
Selected for direct evidence about the subject, its availability or the limits of the claim.
Anthropic · page checked 2026-10-18
Coverage note: Independent hands-on comparisons and current cross-platform discussion are incomplete in this bounded sample. X public search required login.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
Anthropic describes broader access for independent evaluators assessing advanced AI risks, including embedded access for six METR employees.