Anthropic grants METR employee-level model access

External evaluators getting insider access could become a template for third-party AI risk oversight.

Anthropic is giving nonprofit evaluator METR employee-like access to its models so researchers can monitor AI risks from inside, an unusual transparency arrangement for an AI lab.

Read it at CNA →

Coverage: 22 sources, in order

1. 15 Oct 17:35 CNA Read the article →
2. 15 Oct 17:35 CNA Read the article →
3. 15 Oct 17:35 Reuters Read the article →
4. 15 Oct 17:42 Unite.AI Read the article →
5. 15 Oct 17:43 Kansas City Star Read the article →
6. 15 Oct 17:52 Seeking Alpha Read the article →
7. 15 Oct 17:52 TradingView Read the article →
8. 15 Oct 18:25 Daily Sabah Read the article →
9. 15 Oct 18:25 Daily Sabah Read the article →
10. 15 Oct 18:29 Business Insider Read the article →
11. 15 Oct 19:12 Financial Times Read the article →
12. 15 Oct 19:12 Financial Times Read the article →
13. 15 Oct 19:14 TechCrunch Read the article →
14. 15 Oct 19:14 TechCrunch Read the article →
15. 15 Oct 19:22 Benzinga Read the article →
16. 15 Oct 19:30 The Lufkin Daily News Read the article →
17. 15 Oct 19:37 The Information Read the article →
18. 15 Oct 20:08 Reddit Read the article →
19. 15 Oct 20:35 CNBC Read the article →
20. 15 Oct 20:35 CNBC Read the article →
21. 16 Oct 00:08 NewsCord Read the article →
22. 16 Oct 04:02 Briefs Finance Read the article →

← Front page