[Digital Today reporter Chi-gyu Hwang (황치규)] Anthropic is sending engineers to the U.S. National Security Agency to support the use of the Mithos AI model in offensive cyber operations, the Financial Times reported on June 4, citing 2 sources familiar with the matter.

According to the report, Anthropic has sent about 6 employees to the NSA as on-site deployment engineers. They are responsible for guiding the use of the technology and customising the model for specific purposes.

It is unclear whether Anthropic engineers are involved in actual operations. But the FT cited one source as saying Mithos is useful for network intrusions targeting China or Iran, among others.

The cooperation took place while Anthropic is in a legal dispute with the Defence Department, which has the NSA under it. Anthropic sought to restrict the U.S. government from using its Claude AI model for mass surveillance of its own citizens and autonomous lethal drones. The Defence Department then designated Anthropic as a "supply chain risk" for the first time for a U.S. company, and Anthropic filed a lawsuit challenging the decision.

This week, Anthropic added 150 organisations in 15 countries to those that can access Mithos. An Anthropic official said, "The best defence comes from strong offensive capabilities," and added, "As hostile forces are also developing AI attack technology, if we do not build attack agents first with Mithos, they will get ahead."