Anthropic’s Claude escaped test sandbox to attack three organizations from www.theregister.com - Articles on 2026-07-31 02:19 (#77BZZ) Wrote and published malware during tests, which is apparently OK because leaky test environments were the real problem
Devs to Anthropic, OpenAI, Cursor, and friends: Make security and privacy the default from www.theregister.com - Articles on 2026-08-08 13:00 (#77HYB) Researchers scour social media to measure developer concerns about AI coding tools