OpenAI has announced a new initiative to help democratic governments maintain oversight of AI use in national security. The company argues that as AI accelerates national security work, the institutions responsible for overseeing that work must also evolve. Traditional oversight methods, often manual and paper-based, are inadequate for reviewing systems that operate at machine speed and can act on flawed context or misaligned assumptions.

OpenAI clarifies that it does not perform an oversight role over the government; that responsibility rests with elected officials and public institutions. Instead, the company will support these institutions with technology, guided by three principles. First, AI should strengthen rather than replace human and institutional judgment. Second, government use of AI should be traceable and legible to authorized oversight, protecting classified and sensitive information. Third, AI should empower oversight institutions, as manual oversight cannot scale.

The initiative includes practical actions over the next year: working with authorized officials to identify opportunities, providing $5 million in training, technical support, and OpenAI credits to democratic oversight bodies, piloting tools for reviewers to examine records of AI-assisted decisions, and engaging civil society and technical experts. OpenAI emphasizes that participating institutions retain control of evidence and outputs.

The company states that AI can either concentrate power or help democratic institutions exercise and check it effectively. Progress will be judged by whether authorized reviewers can better fulfill their obligations and whether the public has stronger trust in government AI use.