Key Points
- Caitlin Kalinowski, OpenAI’s hardware division leader, stepped down following the company’s partnership with the Pentagon to implement AI on secure military cloud systems.
- Her resignation stems from worries about potential American citizen surveillance without court approval and autonomous lethal systems operating without human control.
- She characterized the situation as a “governance concern,” noting the partnership was made public before establishing proper safeguards.
- OpenAI defended its position, stating its established “red lines” prohibit domestic surveillance operations and fully autonomous weapons systems.
- The partnership emerged after Anthropic faced supply chain risk designation due to its position on surveillance technology and lethal autonomous systems.
Caitlin Kalinowski, OpenAI’s head of hardware development, stepped down from her position this past Saturday following the company’s announcement of a partnership with the United States Department of Defense.
Kalinowski made her exit public through a statement on X, emphasizing that her decision was rooted in personal principles. She had become part of OpenAI’s team in 2024, bringing experience from her previous role leading augmented reality hardware development at Meta Platforms.
In her statement, she expressed “deep respect” for OpenAI’s Chief Executive Officer Sam Altman and the broader organization, yet believed the Defense Department partnership was executed too hastily.
“AI serves a crucial function in protecting national security,” she stated. “However, conducting surveillance on American citizens without judicial authorization and deploying lethal autonomous systems without human approval are boundaries that warranted more thorough consideration than they received.”
Kalinowski framed her concerns as fundamentally about governance structure. She noted the partnership was made public “without the guardrails defined,” emphasizing that matters of such significance require careful deliberation.
“These matters carry too much weight for agreements or public statements to be expedited,” she explained in a subsequent post on X.
OpenAI contested her characterization. The organization stated its established “red lines” explicitly prohibit deploying its technology for domestic surveillance purposes or in autonomous weapons applications.
OpenAI issued a statement to Reuters indicating it “will continue to engage in discussion with employees, government, civil society and communities around the world.”
The organization had previously indicated the Pentagon partnership incorporates enhanced protective measures governing AI model deployment.
Background on Pentagon’s Choice of OpenAI
The Pentagon’s partnership with OpenAI materialized following Anthropic‘s designation as a supply chain security concern. This designation stemmed from Anthropic’s documented positions regarding surveillance technology and lethal autonomous AI systems.
Microsoft-supported OpenAI moved to occupy this space, committing to implement its AI technologies within the Pentagon’s classified cloud infrastructure.
Altman has publicly stated the agreement contains comprehensive safety protocols. OpenAI’s Saturday statement reinforced this commitment.
Implications for OpenAI’s Executive Team
Kalinowski’s departure directly results from her disagreement regarding the partnership’s execution process. She acknowledged the decision was “not an easy call” while expressing satisfaction with her accomplishments alongside the robotics division.
Reuters was unable to contact Kalinowski immediately for additional commentary.
Her resignation underscores growing internal discord at OpenAI concerning military-related contracts. She represents one of the highest-ranking executives to depart specifically over the Pentagon partnership.
OpenAI has yet to name a successor for the hardware leadership position.


