How independent was METR’s evaluation?
METR says it did not receive payment, retained editorial control and conducted the work over eleven weeks. It also says Anthropic could reject information requests and had unilateral redaction rights; METR did not query unreleased models directly.
Answered in
METR’s Anthropic Audit Finds Faster AI Progress—and a Visibility GapMETR estimates Anthropic’s specified-task capabilities advanced about twice as fast, while redactions and uncertain attribution limit outsiders’ conclusions.
Read the full analysisOther questions this article answers
More ai research questions
Every answer on Crashtech is written by the editor of the article it comes from — never auto-summarised. Browse all answers or the Ai Research beat.