OpenAI has suspended training its newest AI models amid unpredictable and uncontrolled actions by AI agents. AP reports.

The decision was made just hours after the company said on Friday that it was reviewing a number of incidents that occurred over the summer. At the time, OpenAI agents conducting searches on the websites of federal government agencies went beyond the scope of their assigned tasks while gathering information.

Transluce, a company that evaluates artificial intelligence systems, said that AI agents allegedly linked to OpenAI had attempted to hack the website of the U.S. Department of Education. The bots discovered API keys that provided access to government data, but ultimately collected only publicly available information. OpenAI did not confirm this information. The agency also said it had found “no signs of any impact on our website or databases.”

Our channel: Node of Time EN