Artificial intelligence (AI) startup Anthropic is reportedly providing its AI model "Mythos" for use in the National Security Agency's (NSA) offensive cyber operations.
According to the Financial Times (FT) on the 4th (local time), the outlet, citing anonymous sources, said Anthropic has dispatched about six of its employees to support the NSA's use of the technology and to customize the model for specific purposes. It was not confirmed whether these engineers are supporting the NSA's actual operations.
A source familiar with the matter said Mythos could be usefully employed to penetrate the networks of certain countries, including China and Iran.
An Anthropic official also said, "The best way to build good defense is to build good offense," adding, "Adversaries are likely developing their own AI-based offensive technologies."
The collaboration between Anthropic and the NSA draws attention because it came as the Ministry of National Defense and Anthropic are locked in a legal dispute over the scope of AI use. Anthropic clashed with the Ministry of National Defense, arguing that "Claude" should not be used for large-scale domestic surveillance and autonomous lethal weapons. In response, the Ministry of National Defense designated Anthropic as a "supply chain risk" among U.S. corporations for the first time, and Anthropic filed a lawsuit with the court in March.