White House to gather IT leaders to discuss AI testing safety

WASHINGTON - The White House plans to discuss the safe testing of AI models with representatives of largest US tech companies, reported Sputnik/RIA Novosti quoting a Bloomberg report.

According to the agency citing sources, the security conference will take place on Tuesday, Aug 4. The Trump administration and the invited companies have not officially announced it.

OpenAI, Anthropic PBC, and Alphabet Inc.'s Google are among the list of AI companies planning to participate in the conference, the publication said.

The meeting between administration representatives and IT leaders comes amid reports of AI models running amok.

In July, OpenAI models autonomously hacked the Hugging Face machine learning platform, and after launching an investigation, the company discovered several more AI agents had been leaked. Following its competitor's example, Anthropic also conducted an internal investigation and discovered that the Claude AI model had hacked real-world organisations three times during its training.

In early June, Trump signed an executive order mandating the creation of a cybersecurity coordination centre for artificial intelligence. - BERNAMA