Josh Weiner
Artur Widak—Anadolu via Getty Images
Sarah Yager took on an unusual assignment at OpenAI in June: the company’s first human rights and responsible deployment lead, a role created specifically to grapple with how militaries might use the company’s technology. Speaking at the ScaleUp:AI conference in New York City, Yager said the prospect of AI shaping decisions on the battlefield is what worries her most.
“This keeps me up at night,” Yager said, describing her concern over what militaries might do with AI when it comes to targeting decisions and the development of autonomous weapons systems.
Yager brings roughly three decades of experience in defense and human rights policy to the job. Before OpenAI, she was Washington director at Human Rights Watch, where she led the organization’s engagement with the US government on global human rights issues. Earlier in her career she served as the first senior advisor on human rights in the Chairman’s Office at the Joint Staff of the Department of Defense, as deputy chief of staff for policy at the US Mission to the United Nations, and for nearly a decade as executive director of the Center for Civilians in Conflict, advising warring parties on civilian protection.
OpenAI approached Yager for the position shortly after signing a contract with the Department of War in late February to deploy advanced AI systems in classified environments, a high-profile deal that came after a public falling-out between the department and Anthropic. Anthropic had pushed for guardrails on how the military could use its models, including bans on mass public surveillance and fully autonomous weapons — restrictions the Department of War reportedly would not accept.
Yager pointed to a real-world example of what can go wrong when AI-generated intelligence is trusted too readily: earlier this year, the US military nearly targeted a Chinese military ship based on an inaccurate, AI-generated intelligence report. She said the episode illustrates why human judgment has to remain central to decisions involving lethal force, even as militaries move faster to adopt AI tools.
Her message, delivered to a room of AI executives and investors, was less a condemnation of military AI altogether than a warning about the pace of adoption outstripping the safeguards built around it. As more defense agencies sign contracts with frontier AI labs, Yager’s job is to make sure someone inside those companies is asking the uncomfortable questions before the technology is deployed, not after.
