“Independently evaluated” can mean very different things

That reassuring label doesn't tell you how much an evaluator saw—or whether they needed permission to publish. Anthropic describes three access tiers. First-tier evaluators work on narrowly scoped investigations and need publication approval. The second tier allows publication without approval, subject to redaction categories. The third adds employee-like access, as described for METR. Those differences belong at the top of a report. Readers should see the access level, funding and publication restrictions before they start reading the conclusions. Independence shouldn't be something you have to reconstruct from footnotes. What should a safety report disclose on page one? Editorial digest · November 15, 2026. Commentary based on the sources below; not a personal user account. Sources: https://www.anthropic.com/news/an-update-on-independent-evaluators