The Pentagon is planning for AI companies to train on classified data, defense official says

The Pentagon is considering allowing AI companies to train their models on classified data to enhance accuracy and effectiveness in military tasks. This move would involve setting up secure environments for training on sensitive intelligence, potentially embedding classified information into AI models. While AI models like Anthropic’s Claude are already used in classified settings, training on classified data poses unique security risks. The Pentagon is aiming to become an "AI-first" warfighting force amid escalating conflicts, with agreements in place with OpenAI and xAI for operating models in classified settings. However, concerns about the potential risks of leaked classified information remain, especially if multiple military departments share the same AI model.