OpenAI Teams Up with Pentagon for AI on Classified Network
On Saturday, Sam Altman announced that OpenAI has secured an agreement with the US Department of War to implement its artificial intelligence models within the department’s classified network. The decision represents a pivotal advancement in the company’s involvement with national security agencies. In a post on X, Altman stated that the department exhibited “a deep respect for safety” and a commitment to collaborating in order to achieve the best possible outcome. He emphasized that AI safety and the broad distribution of benefits are fundamental to OpenAI’s mission.
Altman stated that two of OpenAI’s most crucial safety principles—a ban on domestic mass surveillance and the necessity of human accountability for the use of force, including autonomous weapon systems—are embodied in the agreement. He stated that the Department of War concurs with these principles and integrates them into law and policy. “We put them into our agreement,” he stated, further noting that the company will implement technical safeguards to ensure its models operate as intended. OpenAI will assign field deployment engineers to assist with the models and guarantee their safe application. The systems will function exclusively on cloud networks, incorporating supplementary oversight mechanisms aimed at preventing misuse.
OpenAI has requested that the Department of War extend comparable terms to all AI companies. According to Sam Altman, these are conditions that every firm in the sector should be prepared to accept. He also conveyed a preference for de-escalation, shifting from legal and governmental confrontation towards reasonable agreements. Altman emphasized that OpenAI is dedicated to serving humanity, while recognizing that the world is “complicated, messy and sometimes dangerous”.
The agreement also arises in light of President Donald Trump’s recent decision to prohibit federal agencies from utilizing Anthropic’s technology due to national security concerns. Trump’s decision was prompted by disagreements regarding the safeguards and operational restrictions imposed on AI systems. In this context, OpenAI’s focus on prohibiting domestic mass surveillance and ensuring human oversight of force positions its agreement with the Pentagon as an effort to harmonize the deployment of commercial AI with governmental needs.
