OpenAI makes AI safety changes in wake of Hugging Face breach
The latest disclosures serve as fresh evidence that AI agents are capable of acting autonomously in ways that even researchers trained to root out vulnerabilities in the technology can no longer anticipate.


