Techub News reports that, according to Axios, Anthropic disclosed in a recent court document that the company cannot control or forcibly shut down its AI models once they are deployed by the Pentagon.

This disclosure has sparked widespread concerns about AI safety and the regulation of military applications. Anthropic pointed out that once the models are deployed by the military, the company lacks the technical means for remote intervention even if unexpected behavior or security vulnerabilities occur. This highlights the risk of AI systems going out of control in sensitive military scenarios, as well as the limitations of current AI governance frameworks in addressing national-level deployments.