OpenAI Top Executive Leaves Company Due to Pentagon Deal
Caitlin Kalinowski, who led robotics work at OpenAI, has stepped down. She told colleagues and the press that the break came after she clashed with the company over how fast it moved to finalize a deal with the U.S. Department of Defense.
Her critique was blunt: the contract, she said, was sealed before key safety measures for neural networks were in place. Kalinowski flagged two concerns in particular—possible surveillance of U.S. citizens and the deployment of AI-driven autonomous combat systems—which, she argued, deserved more time and care (e.g., clearer limits and safeguards).
This next phase of OpenAI’s work with the Pentagon followed the Defense Department ending its relationship with Anthropic, a move tied to that company and the Pentagon not finding common ground on usage limits for certain military tasks.
OpenAI told Engadget it does not support the kinds of uses Kalinowski described. Company reps pointed out that talks with the Pentagon included “red lines,” such as bans on surveilling U.S. citizens and on fully autonomous weapons. Still, Sam Altman—the CEO—has cautioned that the current contract text may not spell out those restrictions in full; he suggested the relevant language could be added later.