Media reports, including Bloomberg, indicate that the National Security Agency is using an advanced AI model known as “Claude Mythos,” despite internal efforts within the Pentagon to restrict its deployment.
According to the reports, Jerome Powell and Scott Bessent issued joint warnings to banks about “unprecedented cybersecurity risks” associated with such advanced models, highlighting growing concerns over AI’s impact on financial and national security systems.
Meanwhile, a safety report from Anthropic acknowledged that the model has reached a level of complexity that makes its behavior difficult to fully measure or predict.
These developments underscore a critical paradox: technologies considered too risky for broad public deployment are already being integrated into sensitive national security infrastructure.
As the global AI race accelerates, the debate over balancing innovation with safety continues to intensify, particularly in high-stakes sectors like defense and finance.