The U.S. Department of Defense has classified Anthropic as a supply chain risk, preventing the developer's AI from being used in government applications. However, Mythos, a model focused on cybersecurity, is being closely monitored by the government and may be an exception.
I think the issue of Mythos, which is being addressed across the government, not just the War Department, is a separate national security moment where we need to ensure our networks are hardened, because this model has specific capabilities to find cyber vulnerabilities and fix them.
The assessment was made this Friday (1) by the agency's technology director, Emil Michael, during CNBC's Squawk Box program. According to him, Claude Mythos is a special case, as it has strategic relevance for US national security.
The statement comes after a standoff between the Pentagon and Anthropic, which gained momentum earlier this year. The government wanted unrestricted use of Claude in military applications, while the developer did not authorize its use in tasks considered dangerous, such as citizen surveillance and autonomous weapons. Olhar Digital broke down this dispute here.
Despite restrictions, the US is using Mythos
Despite the formal restrictions, the situation is more complex in practice.
According to CNBC, there are reports of the company's models being used in operations related to the war against Iran, as well as indications that Mythos is being employed by the National Security Agency (NSA). This context raises questions about how the government can use the technology without contradicting its own guidelines.
News item retrieved from the Samir News website.