ทันเอไอ

OpenAI Unveils Security Collaboration Principles, Emphasizes 'Red Lines' Against Autonomous Weapons Use

OpenAI has published guiding principles for collaboration with governments and security agencies, establishing clear restrictions on AI technology's use for military and national defense purposes.

OpenAI has officially unveiled a set of guiding principles for collaborating with governments and national security agencies. Developed in conjunction with expert David Kris, these principles will apply to both current and future partnerships. A central component involves establishing clear 'red lines' through contractual limitations. These prohibit the use of OpenAI's technology for mass domestic surveillance, to command autonomous weapons systems (weapon systems capable of independently deciding to attack targets), and for high-risk automated decision-making. This announcement occurs amid tensions between the U.S. government and rival Anthropic, which declined to remove safety restrictions from its Claude model for military use earlier in 2026. OpenAI has also expanded its cybersecurity cooperation via the Daybreak project with several countries, including Australia, Japan, South Korea, and European Union nations. These principles have already been integrated into existing collaborations with the Department of War, with OpenAI asserting full control over its safety stack.