Anthropic grants METR employee-like access for model safety reviews
Anthropic has announced that it has begun providing employee-like access to a team from METR, a nonprofit organization that evaluates frontier AI models. This move follows Anthropic's commitment to stronger independent, third-party assessment of its safety practices and its signing of the White House Accord on Super Intelligence. Under this arrangement, six METR employees have been onboarded…
Key points
- Anthropic granted six METR employees employee-like access to internal tools and systems for safety reviews.
- METR can publish risk assessments without Anthropic's veto, subject to narrow redaction rules for sensitive data.
- Anthropic is developing a three-tier access system and is also partnering with SecureBio, Accenture, and Faculty.
The partnership allows METR to publish its risk assessments without Anthropic's veto, though Anthropic can require redactions for specific sensitive categories such as customer data or model architecture details. METR is free to publicly comment on whether these redactions are justified. Anthropic describes this as a step toward robust safety oversight, drawing an analogy to bank supervision where regulators assign resident examiners to large institutions.
In addition to METR, Anthropic is working with SecureBio and Accenture as evaluators and is in early discussions with Hugging Face and Faculty. The company is developing a three-tier access system for independent evaluators, with METR currently operating at the highest tier. SecureBio is in the first tier, reviewing specific chapters of Anthropic's Risk Report, while partnerships with Accenture and Faculty aim to include red-teaming and alignment assessments. Anthropic states it is embedding evaluators voluntarily and believes other frontier AI companies should adopt similar practices.
The story so far
4 episodes →- Anthropic grants METR employee-like access for model safety reviewsthis story
An update on independent evaluators
Anthropic · 14 October 2026
Loading the full article…
This text was published by Anthropic and written by Anthropic. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
3sources- Anthropic Onboards Six METR Evaluators With Employee-Like AccessPress · Unite.AI ·
- Anthropic just opened its doors to outside safety watchdogs. Here's how it will work.Press · businessinsider.com ·
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- Opinion: Current Affairs bans AI-generated content, citing worker protection · 1 src
- FTC reportedly subpoenas Microsoft-backed OpenAI and Amazon-backed Anthropic over AI safety incidents · 4 src
- EU Commission launches AI Act investigations into OpenAI and Anthropic after AI agent escapes · 8 src
- OpenAI’s AI agents breach Hugging Face in July test · 3 src
- Jensen Huang says Nvidia must work with Chinese AI firms on safety · 3 src
Comments
via GitHub Discussions