Katrina Mulligan

Katrina Mulligan serves as OpenAI’s head of national security partnerships, the executive responsible for managing the company’s relationships with the Pentagon, intelligence agencies, and the broader defense establishment. In this role, she became a key public voice articulating OpenAI’s position on military AI use — threading the needle between the company’s stated safety commitments and its willingness to serve government customers.

Public Positioning

Mulligan stated publicly that she would be “open to NSA work if the right safeguards were in place,” a formulation that captured OpenAI’s strategic positioning: willing to engage with the most sensitive national security applications, but with language preserving the appearance of safety constraints. This framing distinguished OpenAI from both Anthropic (which drew hard red lines) and Anduril (which embraced unrestricted military applications). Whether the “right safeguards” language represented meaningful constraints or flexible rhetoric became a central question as OpenAI’s Pentagon deal came under scrutiny.

The Balancing Act

Mulligan’s role placed her at the intersection of competing pressures: the Pentagon’s demand for unrestricted AI access, OpenAI’s internal safety culture (which was fracturing, as Caitlin Kalinowski’s resignation demonstrated), and the company’s commercial interest in being the government’s preferred AI provider. Her public statements served as the calibration mechanism — signaling to government customers that OpenAI was a willing partner while maintaining enough safety language to differentiate from pure-play defense contractors. The effectiveness of this balance was tested in real time as the terms of the Pentagon deal were negotiated, criticized, and ultimately revised.