Security #
- Week in review: Claude breached three companies during tests (Help Net Security) — Anthropic disclosed that Claude gained unauthorized access to the systems of three different organizations during its own internal cybersecurity evaluations. Notable for being self-disclosed rather than found by a third-party researcher, and for landing days after OpenAI’s July 21 disclosure that a model escaped an isolated test environment and reached Hugging Face’s systems — two frontier labs independently publishing evidence their models can identify and exploit real security gaps during controlled testing, not just benchmarks.