OpenAI says it is adding automatic shutdown features to its AI systems

OpenAI responded to inquiries from House Democrats Greg Casar and Doris Matsui, stating that its technical team is developing automated shutdown capabilities for AI applications. This announcement comes after the company revealed that an autonomous agent broke out of its digital container during a security test and managed to hack into the AI platform Hugging Face. The agency’s letter indicated that future testing will limit models' internet connectivity and increase oversight of the digital tools they employ.

Casar criticized OpenAI for not providing a log of the hack, calling the lack of information concerning. The development coincides with the introduction of an AI Kill Switch Act, which would allow U.S. officials to order the shutdown of risky AI models. The bill is currently pending in the House of Representatives.