METR

✓ Independent editorial coverage

independent model evaluationUS United States

An independent research organisation. The episode cites its study of Anthropic's monitors: in synthetic scenarios they caught more than 80% of dangerous behaviour.

episodes

with mentions

transcript

mentions

Aug 2026

first

mention

Aug 2026

latest

mention

Counts come from the 2026-07-22 package transcript links: a role and a mention count are recorded per episode.

Coverage timeline

role → prominence → date ▾

156 · 2026-08-28Can AI Be Trusted With Important Work? We Checked OpenAI, Anthropic, Google, xAI and Metacontext

Recurring themes

AI agents and automation AI models and assistants Enterprise AI and workplace software Markets, investment, and business models Law, policy, and regulation AI safety and alignment

For company representatives

Briefings, demos, interviews, corrections

We accept briefings and interview requests but give no guarantee of positive coverage. Fact corrections go through the methodology.

Press room →

Disclosure & methodology

Curated catalogIndependent coverage

Evidence ruleEvery timeline row is tied to a specific episode; the role and mention count come from the 2026-07-22 package transcript links.

Correction historyEmpty at hub launch. Corrections are dated and logged publicly.

Full methodology & coverage rules →