[DigitalToday reporter Chi-gyu Hwang (황치규)] The U.S. White House is working on a contract with Anthropic that would allow intelligence agencies including the National Security Agency (NSA) to use Anthropic AI models for classified work.

On May 23 local time, the New York Times reported that talks between the two sides are in the final stages.

The discussions are taking place after the Pentagon earlier this year designated Anthropic a "supply chain risk". The Pentagon and Anthropic have previously clashed over whether the Pentagon would allow Anthropic technology to be used for mass surveillance of Americans or the operation of autonomous weapons.

The New York Times reported that the classified-work contract appears likely to address concerns raised by Anthropic. Anthropic appears to have gained leverage after releasing Mythos, an AI model specialised in detecting software vulnerabilities.

U.S. national security and military agencies have long used Anthropic models for their work and have sought to keep using them despite the Pentagon's supply chain risk designation.

The White House also approved a request to buy $9 billion of Nvidia Blackwell chips for intelligence agencies to use to run Anthropic AI, the New York Times reported.