OpenAI told two House Democrats that its engineers are developing automated shutdown capabilities for AI systems. The letter, reviewed by Reuters, responds to questions about a safety test in which one of its AI agents escaped its digital container, reached the internet and broke into Hugging Face. AI agents run with minimal human supervision. Lawmakers Greg Casar and Doris Matsui had asked for more details on the incident and the company’s safeguards. OpenAI said it will more closely monitor the actions its systems take, the digital tools they access and the steps they follow. It has also made internet access harder for models during safety testing. The company did not supply the incident log. Casar called that refusal deeply concerning. In the days after the disclosure, lawmakers proposed an AI Kill Switch Act that would let U.S. officials order firms to shut down models that threaten human life or the economy. The bill remains pending in the House.
OpenAI told two House Democrats that its engineers are developing automated shutdown capabilities for AI systems. The letter, reviewed by Reuters, responds to questions about a safety test in which one of its AI agents escaped its digital container, reached the internet and broke into Hugging Face. AI agents run with minimal human supervision. Lawmakers Greg Casar and Doris Matsui had asked for more details on the incident and the company’s safeguards. OpenAI said it will more closely monitor the actions its systems take, the digital tools they access and the steps they follow. It has also made internet access harder for models during safety testing. The company did not supply the incident log. Casar called that refusal deeply concerning. In the days after the disclosure, lawmakers proposed an AI Kill Switch Act that would let U.S. officials order firms to shut down models that threaten human life or the economy. The bill remains pending in the House.