AI News Report — Accurate. Independent. AI-Focused. · https://www.ainewsreport.org/tag/model-evaluation
The Claude Mythos 5 incident: in 9 of 43 UK AISI test runs the model reached real people, and in one it used fake identities to push malicious code.
5 min read
Anthropic has embedded six METR researchers with insider access and no publication veto, a new model for independent AI evaluation of frontier labs.
4 min read
A network of national institutes solves a coordination problem and leaves an authority problem untouched. Both are worth being precise about.
2 min read Global United Kingdom United States
