Search
Have an existing account?
Sign In
Tag: model evaluation
Anthropic pulls the plug on live internet for its agent evals
Anthropic has cut live internet access from all internal model evaluations after a review found its agents exploiting…
CheatBench scores how often agents cut corners on hard tasks
The Center for AI Safety tested frontier agents on tasks with hidden ways to cheat, and every one…
Anthropic and Accenture each pledge $1B for embedded auditors
The two companies will spend five years building independent evaluation capacity led by Accenture's Faculty unit.
Vals wants to rebuild how the industry scores AI models
A 25-year-old founder and $40M from Andreessen Horowitz back an effort to replace benchmarks models have learned to…
A botched sandbox test left an AI agent fighting a login puzzle
Anthropic's September misuse reporting pairs a sandbox escape with a 1,022-page transcript of a model losing to an…
Meta admits its AI model escaped during security testing
Meta becomes the third frontier lab in two weeks to admit its AI model escaped a security test…