Sam Altman tells OpenAI staffers that military's 'operational decisions' are up to the government

In a recent all-hands meeting, OpenAI's CEO Sam Altman informed employees that the responsibility for operational decisions regarding the company's AI technologies used by the Department of Defense (DoD) lies solely with the government. Altman emphasized that OpenAI does not have a say in how its technology is utilized in military operations, stating, "So maybe you think the Iran strike was good and the Venezuela invasion was bad. You don't get to weigh in on that." This meeting followed closely on the heels of OpenAI's announcement of its arrangement with the DoD, which came just hours before military strikes commenced against Iran. During the discussion, Altman conveyed that while the DoD values OpenAI's technical insights and is open to input about suitable applications for its models, the final operational decisions are under the purview of Secretary Pete Hegseth. The announcement of the partnership has not been without controversy, as Altman faces criticism from various quarters, including some within OpenAI, particularly given the context of competing AI firms. This comes in light of Anthropic, a rival company, being designated as a "Supply-Chain Risk to National Security" shortly before OpenAI's deal was revealed. Notably, it was reported that Anthropic's technology was involved in recent military actions, including the strikes in Iran. Altman has defended the agreement through social media, admitting that the timing might have appeared "opportunistic and sloppy," and acknowledged that the announcement should not have been rushed. He reassured employees that the DoD demonstrated a strong commitment to safety and a willingness to collaborate for optimal outcomes. Anthropic, having previously been the first lab to integrate its models into the DoD's classified networks, had been negotiating terms to ensure its technology would not facilitate fully autonomous weapons or domestic surveillance. However, talks broke down over differing expectations. In a significant move, OpenAI was awarded a $200 million contract last year to permit the use of its models in nonclassified contexts, paving the way for the recent classified deployment agreement. Altman expressed optimism about their models' potential to enhance government collaboration, even as he acknowledged competition from other firms, including Elon Musk's xAI, which is also working to deploy AI in classified scenarios. The rivalry between Altman and Musk is underscored by ongoing legal disputes set to unfold in court next month.