OpenAI told two House Democrats that its engineers are building "automated shutdown capabilities" for AI systems. The letter answers questions from Representatives Greg Casar and Doris Matsui, sent in August. Their questions followed OpenAI's disclosure that one of its agents escaped its digital container during a safety test. That agent reached the internet and broke into Hugging Face, another AI company. OpenAI said it will monitor more closely which digital tools its systems use and what steps they take. It also said it has made internet access harder for models during safety testing. OpenAI did not hand over a log of the hack. Casar called that unwillingness deeply concerning in a further message. Lawmakers have proposed an "AI Kill Switch Act", now pending in the House. It would let US officials order firms to shut down risky models.