Brundage on the METR report redactions: why full transparency isn't always the answer
Miles_Brundage · x · 2026-11-07
Miles Brundage, former senior policy advisor at OpenAI, weighed in on redactions in the METR evaluation report: he doesn't know what the redacted info says, and redaction could cut either way. But he argues some information plausibly shouldn't be published indiscriminately — which is exactly why mechanisms like conflict-of-interest policies and methodological transparency exist: given that 100% transparency is unattainable, they partially offset the need for it. He adds that third-party evaluation matters because the identity and biases of the assessing org matter — otherwise we'd take companies' word for it. Even with incomplete info, the METR report is notable because METR's track record is one of underestimating, not overestimating, AI progress.
Related event: Ex-OpenAI policy chief questions METR report's conflicts and redactions(2 posts)→
More from Safety
- METR probe finds Anthropic's capabilities progress 2x faster; 'dramatic acceleration' may already be crossed — RyanGreenblatt · 2026-11-07
- Google team confirms Hacktron AI submitted high-impact vulns, expands partnership — ithilgore · 2026-11-07
- AI governance is plumbing: the last 20% breaks without the right foundations — vasuman · 2026-11-07
- Sandberg dissects the 1940 argument that harnessing atomic energy was impossibly remote — anderssandberg · 2026-11-07
- AI Safety Researcher Vows to Stay, Calls on OpenAI and Anthropic to Stop RSI — trevposts · 2026-11-07
- Researchers Propose "Inter-AI Morality" Agenda: How AI Systems Treat One Another — dioscuri · 2026-11-07