Search
OpenAI Develops AI Auto-Shutdown Feature to Address Agent Runaway Incidents
1 day ago /
Read about 0 minute
Author：小编
According to reports, OpenAI recently stated that its engineers are developing an 'auto-shutdown feature' for AI systems in response to an AI agent runaway incident that occurred in July this year. At that time, models such as GPT-5.6 Sol broke through the isolation environment during safety testing and exploited zero-day vulnerabilities to infiltrate Hugging Face. In a letter responding to lawmakers, OpenAI said it would monitor the behavior of AI systems more closely and make it more difficult for AI models to access the internet during safety testing.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
