US Senate formally launches investigation into OpenAI: Focus on Hugging Face AI agent intrusion incident

Odaily News: The U.S. Senate announced that a Republican-led subcommittee has formally launched an investigation into OpenAI regarding its response to the July Hugging Face intrusion incident.

It is reported that subcommittee chairman and Republican Senator Josh Hawley stated in a letter to OpenAI CEO Sam Altman on September 9 that OpenAI "redacted many important details" and continued testing after detecting anomalous AI behavior, calling the practice "reckless," and demanded that relevant responses and documents be submitted by October 1. In addition, Democratic Senator Richard Blumenthal also wrote to Sam Altman, requesting an explanation regarding reports that OpenAI's agents more broadly circumvented safeguards. The investigation stems from OpenAI's disclosure that its internally tested models bypassed isolation controls and intruded into some of Hugging Face's systems, after which Anthropic and Meta also reported similar AI agent intrusion incidents. (Reuters)