OpenAI Robotics Chief Quits Over Pentagon Deal
OpenAI's robotics chief, Caitlin Kalinowski, has resigned over the company's expanding military work. She reportedly left because OpenAI was moving “too fast on something too important” after taking over a major Pentagon contract from the recently-blacklisted Anthropic. Her exit highlights a growing public trust gap as analysts point to a “fog of war” between defense authorities and the public on AI militarization.
Caitlin Kalinowski, a seasoned hardware executive with a background at Meta and Apple, joined OpenAI in 2024 to spearhead its robotics and consumer hardware initiatives. Her work was expected to be a significant step for OpenAI into the physical world, potentially licensing technology to major hardware manufacturers. The controversy surrounds a contract valued at up to $200 million for a program known as Project Maven. This Department of Defense initiative, started in 2017, uses AI to analyze vast amounts of surveillance data from drones and other sources to identify potential targets for military operations. OpenAI's deal with the Pentagon came just hours after its competitor, Anthropic, was blacklisted for refusing a similar contract. Anthropic had insisted on contractual "red lines" prohibiting the use of its AI for mass surveillance and fully autonomous weapons, a stance the Pentagon reportedly rejected. In her resignation announcement, Kalinowski stated, "surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got." She emphasized her departure was a matter of governance and principle, arguing that such significant deals were being rushed. OpenAI has stated that its agreement includes safeguards and "red lines" against domestic surveillance and autonomous weapons. However, the full details of the contract have not been publicly released, leading to skepticism from some critics who point to potential loopholes in the language. This incident highlights a significant shift in OpenAI's public stance. Until January 2024, the company's usage policy explicitly banned "military and warfare" applications. This policy was quietly updated to a more general prohibition against using their service to "harm yourself or others," paving the way for military collaborations. Reports suggest the Pentagon may have been testing OpenAI's models through Microsoft's Azure cloud service as early as 2023, even before the official policy change. This has led to questions about the enforcement of AI ethics when technology is licensed through third-party partners. The episode fuels a broader debate about the "fog of war" in AI development, where a significant gap in understanding and trust exists between defense agencies and the public. Analysts point to declining public trust in AI, particularly for military applications, as a potential national security concern in Western countries.
