OpenAI’s head of robotics and consumer devices, Caitlin Kalinowski, said on Saturday that she had resigned, days after the company struck a controversial deal with the US Defense Department to deploy AI models on classified military networks.
In a post on X, Kalinowski said, “I resigned from OpenAI. I care deeply about the robotics team and the work we built together. This wasn’t an easy call.”
Kalinowski said AI had a role in national security, but said surveillance of Americans without court oversight and lethal autonomy without human authorization were “lines that deserved more deliberation than they got.” She said her resignation was “about principle, not people,” adding that she had deep respect for Altman and the team and was proud of what they had built together.
Kalinowski, who previously held senior roles at Apple and Meta, led OpenAI’s push to move its technology into physical systems. During her 16 months there, she helped build its “physical AI” program, leading a San Francisco lab of about 100 people training robotic arms to perform household tasks.
The resignation comes at a sensitive moment in the defense AI market. Anthropic had previously been the only company authorized to work on the Pentagon’s classified networks, under a $200 million contract awarded in July 2025.
But Anthropic, the developer of the Claude AI model, entered tense negotiations with the Defense Department recently because it objected to using its models for broad domestic surveillance or fully autonomous weapons.
When those talks collapsed, President Donald Trump directed federal agencies to stop using Anthropic’s technology, before OpenAI announced hours later that it had reached its own deal with the Pentagon.
Sam Altman, OpenAI’s chief executive, had previously acknowledged that the Pentagon deal moved quickly and triggered a fierce backlash.
Even so, the company defended its position after confirming the resignation, saying the deal opens a practical path for responsible uses of AI in national security while keeping clear limits against domestic surveillance and autonomous weapons. It added that it would continue discussions on the matter with employees, the government, civil society, and other communities around the world.
Amid that intensifying dispute, Claude jumped to the top of the free apps chart in Apple’s US App Store at the end of February, overtaking ChatGPT, in a rise linked directly to the debate over the limits of AI use in Pentagon operations and weapons programs.
According to data from Sensor Tower, one of the world’s leading app-ranking firms, Claude began the year outside the US top 100 but rapidly climbed the charts, reaching No. 6 on Wednesday, Feb. 25, and then No. 1 on Saturday, Feb. 28.
Anthropic said its free user base grew 60%, while paid subscriptions doubled, posting daily record numbers.
