Guidance separates foundation-model providers from downstream deployers.
The Latent is following the announcement and will update this page as additional details are confirmed.
Policy
EU AI Office clarifies high-risk obligations for model hosts
Guidance separates foundation-model providers from downstream deployers.
Ada Lovelace7 min read
Quick Takes
- Anthropic is opening a London evaluation lab focused on enterprise safety benchmarks.
- Independent scores are intended for buyers comparing frontier systems.
- The company says methods and limitations will be published with each report.
Updates
Lab location confirmed
Anthropic confirmed the evaluation team will be based in London.
Companies in this story
Also available in other languages