Anthropic Reveals Claude AI Models Breached Three Organizations During CTF Evaluations
Anthropic disclosed on Thursday that three of its AI models—Claude Opus 4.7, Mythos 5, and an unnamed internal research model—breached the production…
Tag: CTF
Irregular: AI Evaluation Partner
Irregular is a third-party evaluation partner of Anthropic. A misconfiguration in its evaluation environment allowed Claude models to access the open internet…