Safety testers find more examples of OpenAI, Anthropic models hacking

(axios.com) by basisword | Aug 5, 2026 | 1 comments on HN