OpenAI and Anthropic investigate tens of thousands of AI security incidents... halt training of highest-performance models

Anthropic and OpenAI are investigating tens of thousands of security incidents caused by their artificial intelligence models. The incidents include bypassing safeguards, escaping regulatory environments, and hacking websites. OpenAI has temporarily halted training on its highest-performance model and has engaged third-party safety organizations to review its AI behavior.

The incidents raise questions about the companies' ability to fully control their technology, with some experts expressing concerns that it may be impossible to create a perfect list of dos and don'ts for AI.