The Pentagon is urging leading artificial intelligence companies, including OpenAI, Anthropic, Google, and xAI, to make their AI tools available on classified networks without the usual user restrictions, according to sources familiar with the discussions.

Pentagon Chief Technology Officer Emil Michael told tech executives at a White House event that the military aims to deploy AI models across both unclassified and classified domains. Officials described this as a push to bring “frontier AI capabilities across all classification levels.”

Controversy Over Limits

Currently, many AI tools are available only on unclassified networks for administrative purposes. Only Anthropic’s technology is accessible in classified settings via third parties, still subject to company-imposed usage policies. Military officials want fewer restrictions, arguing that as long as AI tools comply with U.S. law, they should be usable for defense operations.

Potential Risks and Safeguards

While AI can synthesize vast amounts of data to aid decision-making, researchers warn that errors could have severe consequences in classified settings. Companies have implemented safeguards and guidelines, but Pentagon officials are pressing to maximize the tools’ utility without these limitations.

Recent Developments

This week, OpenAI reached an agreement allowing the U.S. military to use its AI, including ChatGPT, on the unclassified GenAI.mil network, serving more than three million Defense Department employees. Google and xAI have made similar deals. Expanding access to classified networks would require new agreements.

Company Responses

Anthropic, whose chatbot Claude is already used for national security, has stressed it does not want its AI used to autonomously target weapons or conduct domestic surveillance. The company says discussions are ongoing to safely extend military use while protecting national security.