OpenAI publishes national security partnership principles

OpenAI has published new National Security Principles to explain how it handles partnerships with governments and national security organizations. The company said the principles are meant to guide use of its AI systems in sensitive areas such as cyber defense, biosecurity, law enforcement, and public services. It said these uses should preserve human judgment, democratic accountability, and the rule of law. OpenAI also described recent partnerships with the U.S. government and allied countries, including cyber defense and biodefense efforts. It said its existing restrictions prohibit uses such as mass domestic surveillance, autonomous weapons direction, and highstakes automated decisions, and that it supports legislation creating safeguards for highrisk military AI applications.