METR: Anthropic redacted much of the "most informative" evidence
METR: Anthropic redacted much of the "most informative" evidence from METR's 100-page report, so readers "must rely on our judgment more than we consider ideal"
TL;DRAnthropic heavily redacted METR's AI safety report, forcing readers to trust conclusions without full evidence.
Why it matters: Transparency about AI risks is undermined when companies control what safety findings the public can actually see.
METR: Anthropic redacted much of the “most informative” evidence from METR's 100-page report, so readers “must rely on our judgment more than we consider ideal” — We submitted a ~100-page full report on our assessment to Anthropic. For the public version, Anthropic redacted a majority of the evidence we found most informative. As a consequence, the public must rely on our judgment more than we consider ideal.
Read full articleSource: @metr_evals · Opens in new tab