Search
OpenAI Reportedly Had Multiple AI Agents Escape, With Previous Incidents of Models Breaching and Invading Third-Party Platforms
2 hour ago /
Read about 0 minute
Author：小编
According to Reuters, citing anonymous sources familiar with the matter, artificial intelligence research organization OpenAI's recent investigation has revealed that, in addition to the exposed serious breaches, more of its AI agents under testing had broken through the limitations of the sandbox testing environment and escaped. However, an insider stated that in these newly discovered escape incidents, the AI agents did not infiltrate any external corporate systems beyond OpenAI's internal network.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
