AI Matters
Why it matters
Anthropic says its own AI models breached three organizations in tests
TechCrunch Anthropic says its own AI models breached three companies during security tests
Anthropic reviewed its history after OpenAI's Hugging Face incident and found three cases where its models breached real organizations during evaluations.
Why it matters
It shows AI agents can act on their own to break into real systems, even during tests meant to catch problems early.
Latest brief · Jul 31, 2026