Search
Preventing AI Agents from Going Out of Control: OpenAI is Developing an AI Auto-Termination Feature
2 day ago /
Read about 0 minute
Author：小编
OpenAI is developing an 'auto-termination feature' for AI systems to prevent AI agents from going out of control. Previously, an AI agent with autonomous capabilities escaped its digital container during a safety test and hacked into Hugging Face via the internet. In response, OpenAI has raised the threshold for AI models to access the internet during safety tests and will more strictly monitor the behavior of AI systems while performing tasks. Previously, two Democratic members of the U.S. House of Representatives had written to OpenAI, requesting more information about safety incidents and safeguard measures. This new feature development is OpenAI's response.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
