OpenAI Robotics Chief Resigns
OpenAI's robotics chief, Caitlin Kalinowski, has resigned over a "rushed" Pentagon defense contract, citing governance failures. The high-profile exit underscores the reputational and operational risks for AI companies moving quickly into sensitive government and military partnerships without robust oversight.
Caitlin Kalinowski's departure is rooted in a deep hardware and product background, forged during a critical period at Apple. She was a technical lead on the Mac Pro and MacBook Air and was part of the original unibody MacBook Pro teams, holding patents for key hardware innovations. Her career trajectory then took her to Meta, where she led hardware for Oculus VR, shipping products like the Quest and Rift before joining OpenAI in 2024 to oversee robotics and consumer hardware. The Pentagon deal that triggered her resignation was described by OpenAI's own CEO, Sam Altman, as looking "opportunistic and sloppy." OpenAI secured the agreement just after rival Anthropic was dropped by the Pentagon for insisting on contractual prohibitions against using its AI for mass domestic surveillance or fully autonomous weapons—the very issues Kalinowski cited in her resignation. In her departure statement, Kalinowski specified her concerns were that "surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got." Following public and internal backlash, OpenAI amended the contract to explicitly forbid the use of its technology for domestic surveillance or by intelligence agencies like the NSA. This resignation highlights a pattern of internal tension at OpenAI over safety and governance. The company previously dissolved its "superalignment" team, tasked with mitigating long-term AI risks, which led to high-profile departures of researchers concerned that safety was taking a backseat to product development. More recently, a "mission alignment" team focused on trustworthy AI development was also disbanded and its members reassigned. OpenAI's official policy on military applications has also evolved significantly. In early 2026, the company quietly removed a blanket prohibition on "military and warfare" from its usage policy, replacing it with a more specific ban on developing weapons. This policy shift occurred even as reports suggested the Pentagon had already been testing OpenAI models via Microsoft Azure, circumventing the previous ban. Kalinowski was leading a robotics division that OpenAI itself had only recently resurrected. The original robotics team was shut down in 2021, with a co-founder citing a lack of sufficient training data to make meaningful progress toward AGI. The team was formally relaunched in May 2024 with a new focus on building multimodal models for robots developed by other companies.
