Trump administration to step up oversight of closed AI models — NYT

The administration of US President Donald Trump will introduce regular security reviews of closed-source artificial intelligence models to identify potential vulnerabilities and security risks, The New York Times (NYT) reports, citing sources.

The decision followed a meeting between administration officials and executives from leading US technology companies, according to the newspaper.

Unlike open-source AI systems, closed-source models developed by companies such as Anthropic and OpenAI, along with other proprietary AI developers, will be subject to the reviews.

The move also reflects the White House's effort to tighten oversight of advanced AI technologies, whose widespread deployment could pose national security risks, the report said.

Excluding open-source models could benefit companies seeking to catch up with Anthropic and OpenAI, widely regarded as leaders in next-generation AI development. Companies such as Meta have relied on open-source software to build products that compete directly with the two frontrunners, the NYT noted.

Details of the bodies responsible for conducting the reviews have not been disclosed, and the assessments are expected to take place behind closed doors.

In June, Trump signed an executive order establishing a national coordination centre for artificial intelligence cybersecurity in the United States.

By Jeyhun Aghazada