Skip to main content
See every side of every news story
Published • loading... • Updated

Anthropic grants AI evaluator METR employee-level access to its models

Six METR evaluators will assess Anthropic’s models and safety practices with employee-like access, and the nonprofit can publish findings with limited redactions.

  • On Thursday, Anthropic granted six researchers from METR 'employee-like access' to its systems, allowing the nonprofit to independently assess the company's advanced models and safety practices.
  • CEO Dario Amodei's proposal to embed outside evaluators last month prompted this move, positioning METR as the first partner in Anthropic's new three-tier evaluation framework.
  • The researchers received company hardware, office space, and internal tools 'roughly on par' with Anthropic's own risk assessment teams, and can publish findings without prior approval subject to limited redactions.
  • Anthropic is the first frontier AI company to provide such access, aiming to 'close the gap between what evaluators understand' about its systems and what the company itself understands.
  • While Anthropic is also working toward an 'ambitious partnership' with Accenture, the move fulfills a voluntary accord presented by President Donald Trump in late September.
Insights by Ground AI
Podcasts & Opinions

45 Articles

Lean Right

Anthropic AI will be evaluated to review the risks of its models as a firm commitment to be safer in the face of the obvious concern that its technology can be uncontrollable

·Madrid, Spain
Read Full Article
Ars TechnicaArs Technica
Reposted by
MASSIVE.NEWSMASSIVE.NEWS
Center

Anthropic gives outside AI watchdog “employee-like” access

The outside watchdog can publish its findings without Anthropic's approval.

·Cambridge, United States
Read Full Article
Lean Right

On Thursday, the AI developer announced that the non-profit company "Model Evaluation and Threat Research" (METR) had access at employee level. Most recently, there were several security concerns regarding the AI chatbot Claude.

·Düsseldorf, Germany
Read Full Article
Think freely.Subscribe and get full access to Ground NewsSubscriptions start at $9.99/yearSubscribe

Bias Distribution

  • 43% of the sources are Center
43% Center

Factuality Info Icon

To view factuality data please Upgrade to Premium

Ownership

Info Icon

To view ownership data please Upgrade to Vantage

Crypto Briefing broke the news on Thursday, October 15, 2026.
Too Big Arrow Icon
Sources are mostly out of (0)

Similar News Topics

News
Feed Dots Icon
For You
Search Icon
Search
Blindspot LogoBlindspotLocal