# Anthropic commits to embedded third-party AI safety evaluators

_Published Saturday, September 12, 2026 at 11:18 AM EDT · AI, Policy · Latest · Tier 1 — Major_

![Anthropic commits to embedded third-party AI safety evaluators — Primary](https://cdn.prod.website-files.com/67ecbba31246a69e485fdd4b/6aa5311897ee19af5d9f2cd9_og_we-must-pace-the-frontier.jpg)

Anthropic CEO Dario Amodei said the company is unilaterally committing to give embedded third-party evaluators ongoing, employee-like access to assess adherence to safety practices, report incidents and review models, training pipelines and processes. In an essay calling for a slower pace of frontier capability advances, he proposed broader coordination among companies and governments. The company said its own commitment is the first step; industry-wide and global coordination would require others to participate.

## Sources

- [darioamodei.com](https://darioamodei.com/post/we-must-pace-the-frontier)

---
Canonical: https://techandbusiness.org/newswire/eXbCjuc5VXVXwkesPBYCNg
Published: 2026-09-12T15:18:29.316Z
Story chronology: 2026-09-12T14:10:49.000Z
Retrieved: 2026-10-27T20:42:41.253Z
Publisher: Tech & Business (techandbusiness.org)