OpenAI is publishing its National Security Principles to guide the use of its AI technology in sensitive government applications, particularly in national security and law enforcement. The company believes democratic societies should leverage AI for defense, public services, and threat response, citing cyber defense and biological security as areas where AI offers significant advantages.

These principles emphasize deployment that upholds democratic accountability, human judgment, and the rule of law, preventing power concentration. OpenAI has engaged experts like David Kris and internal teams to develop these guidelines. The company is expanding its work with governmental partners, having established Trusted Access for Cyber partnerships with nations including Australia, Canada, and Japan, and collaborated with the UK government on cyber testing.

OpenAI also announced expanded trusted access to its GPT‑Rosalind model for U.S. and allied partners in public health and biodefense. Existing contractual restrictions, such as prohibiting AI use for mass domestic surveillance or autonomous weapons, remain in place and align with these new principles. OpenAI advocates for legislative safeguards on high-risk military AI uses, asserting its role is to inform democratic decision-making, not to dictate policy.