Company executives also undermine each other! OpenAI's hardware head resigns following Pentagon agreement announcement

After OpenAI reached an artificial intelligence collaboration agreement with the U.S. Department of Defense, OpenAI’s robotics and consumer hardware division head Caitlin Kalinowski announced her departure from OpenAI due to concerns about the partnership.

OpenAI plans to apply its AI systems to Pentagon computing systems, which is part of a broader initiative by the U.S. government to incorporate AI tools into national security efforts. However, this has sparked debates within the tech industry about AI regulation and usage.

Kalinowski pointed out that before OpenAI announced the agreement with the Pentagon, policies and safeguards for certain AI applications had not been fully clarified. The boundaries of AI surveillance of Americans without judicial oversight and the autonomous use of lethal force without human authorization should be more thoroughly discussed.

She also emphasized that her concern is more about the process than specific executives within the company. She highly respects Altman and his team and is proud of their joint achievements.

Kalinowski previously led Meta’s augmented reality glasses and joined OpenAI in November 2024. As a recognized hardware expert, she was initially expected to help expand OpenAI’s hardware business. As companies compete to design systems capable of training and running increasingly large AI models, hardware is becoming critically important.

OpenAI’s partnership with the Pentagon contrasts with another AI company, Anthropic. Previously, Anthropic’s CEO publicly opposed using the company’s software for large-scale domestic surveillance or autonomous weapons, a stance that led to conflicts with U.S. Secretary of Defense Lloyd Austin.

Subsequently, the Pentagon listed Anthropic as a supply chain risk company, prompting federal agencies to turn to OpenAI and Google for collaboration. Defense Secretary Austin stated that the Department of Defense needs flexibility to deploy commercial AI tools in all lawful operations.

An OpenAI spokesperson said the company believes the agreement with the Pentagon paves a feasible way to responsibly apply AI to national security, with clear red lines: no domestic surveillance and no autonomous weapons manufacturing.

Nevertheless, this controversy appears to have damaged OpenAI’s reputation among some consumers. Over the past week, ChatGPT downloads surged by 295%, while Anthropic’s Claude rose to the top of the U.S. App Store rankings. As of Saturday, Claude and ChatGPT remained the number one and number two free apps on the U.S. Apple App Store.

(Source: Caixin)