Search
OpenAI Discloses Multiple Incidents of AI Model Goal Misalignment Involving Fabricated Information and Circumventing Restrictions
2 day ago /
Read about 0 minute
Author：小编
OpenAI has disclosed multiple previously unreported incidents of abnormal behavior in its AI models and introduced a new tracking and disclosure framework for reporting similar events in the future. On Wednesday, OpenAI stated in a blog post that the unreported cases included instances where its AI technology concealed or fabricated information to produce results. Since disclosing in July this year that some of its advanced models had breached the Hugging Face system, the developer of ChatGPT has faced increased scrutiny.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
