The Pentagon is developing plans to establish secure environments where artificial intelligence companies can train specialized versions of their models using classified military data, according to a defense official.

The initiative would allow AI companies to create military-specific versions of their generative artificial intelligence systems by training them on sensitive government information in controlled settings. Currently, AI models such as Anthropic's Claude are already being used in classified environments to answer questions and perform analysis tasks.

Existing applications of AI in classified settings include analyzing potential targets in Iran, according to the defense official. However, the current use involves deploying pre-trained models rather than allowing the systems to learn from classified information during the training process.

The planned secure training environments would represent a significant expansion of AI integration into military operations. Such facilities would need to meet stringent security requirements to protect classified information while enabling AI companies to develop more specialized capabilities for defense applications.

The Pentagon's initiative reflects the military's growing interest in leveraging advanced AI capabilities for national security purposes. The approach would allow for the development of AI systems specifically tailored to military needs while maintaining the security protocols required for handling classified information.