AI brief
Anthropic claimed its model Claude Mythos outperforms most security experts at finding software vulnerabilities, and an OpenAI–Hugging Face hacking incident followed.
Written by AI from MIT Technology Review AI's published text. Read the original for full details.