On Saturday, March 7, 2026, Caitlin Kalinowski, OpenAI’s Head of Robotics and Consumer Hardware, announced her resignation in a high-profile move that has intensified the debate over the ethical boundaries of military AI.
Her departure is a direct response to OpenAI’s controversial deal with the U.S. Department of Defense (Pentagon), which was signed just days earlier.
The Breaking Point: “Principle, Not People”
Kalinowski, a veteran hardware leader who previously headed Meta’s AR glasses division and Oculus VR, stated that her decision was rooted in a fundamental disagreement over governance and the speed at which the deal was finalized.
- The Core Concerns: In her announcement on X, she identified two specific “red lines” that she believes the company crossed without adequate deliberation:
- Surveillance of Americans: The potential for AI to be used for domestic surveillance without judicial oversight.
- Lethal Autonomy: The development or deployment of lethal autonomous weapons systems that operate without human authorization.
- The “Rushed” Announcement: Kalinowski emphasized that while AI has a role in national security, the partnership was “rushed without the guardrails defined.” She described it as a “governance concern first and foremost.”
Context: The Anthropic-OpenAI Tug of War
The resignation follows a chaotic week in the AI industry that saw OpenAI’s main rival, Anthropic, take the opposite path.
| Event | OpenAI Response | Anthropic Response |
| Pentagon Terms | Accepted the deal to deploy models on classified networks. | Rejected the deal, citing concerns over surveillance and lethal autonomy. |
| Government Reaction | Praised by the administration for “de-escalating” tensions. | Designated a “Supply Chain Risk” by the Department of Defense. |
| User Reaction | Facing a massive boycott; 1.5 million subscribers reportedly cancelled. | Saw its app climb to the top of the App Store in a wave of public support. |
OpenAI’s Response & Damage Control
CEO Sam Altman has been in active “damage control” mode following the backlash and the loss of key talent like Kalinowski and Max Schwarzer (VP of Research), who also recently left for Anthropic.
- “Opportunistic and Sloppy”: Altman admitted in a post on X that rushing the announcement on a Friday was a mistake that looked “opportunistic.”
- Amending the Contract: OpenAI has moved to update the language of its Pentagon agreement to explicitly prohibit the use of its systems for “domestic surveillance of US persons and nationals.”
- Red Lines: The company maintains that its internal safety stack remains in place and that it will not allow its technology to be used for autonomous “kill decisions.”
What This Means for OpenAI Robotics
Kalinowski’s exit is a significant blow to OpenAI’s physical AI ambitions. She was a key architect in the company’s efforts to bring its models into the physical world through hardware partnerships (such as the rumored collaboration with Jony Ive) and the internal robotics team. Her departure leaves a leadership vacuum at a time when OpenAI is under immense pressure to prove that its “Superintelligence” can safely interact with the real world.
