Anthropic chief executive Dario Amodei has sharply criticized rival OpenAI over its recent agreement with the U.S. government, accusing the company and its CEO Sam Altman of misleading the public about the nature of the deal. In an internal memo to staff, Amodei reportedly described OpenAI’s messaging around its Pentagon partnership as “straight up lies” and accused Altman of “gaslighting” the public about the arrangement.
The dispute centers on OpenAI’s newly announced deal with the U.S. Department of Defense to power certain classified government work. The announcement came shortly after Anthropic was reportedly blacklisted from similar government engagements, intensifying competition between the two artificial intelligence firms. According to reports, the internal memo criticizing OpenAI was first revealed by The Information.
Amodei argued that OpenAI’s claims about stronger safeguards surrounding its defense partnership were misleading. He labeled the company’s handling of the deal as “safety theater,” suggesting that the protections being promoted publicly were more about optics than genuine risk prevention. According to the memo, Anthropic declined a comparable opportunity because it prioritized preventing potential misuse of AI technologies rather than appeasing internal or external stakeholders.
The Anthropic CEO also accused Altman of attempting to present himself publicly as a mediator and responsible dealmaker while masking the true motivations behind OpenAI’s agreement with the U.S. military. Amodei wrote that OpenAI accepted the deal largely to placate employees, whereas Anthropic chose to reject similar arrangements due to concerns about possible abuses of advanced AI systems.
The criticism comes as Altman has spent several days addressing questions about the Pentagon partnership. He acknowledged that the timing of the announcement appeared “opportunistic and sloppy,” particularly because it was revealed shortly after Anthropic’s reported exclusion from similar government work. Altman later expressed regret over how the announcement was handled.
The controversy has also spilled into the public sphere, where reactions from users have intensified competition between the two AI platforms. Following the defense deal announcement, some users of OpenAI’s ChatGPT reportedly threatened to cancel their subscriptions and shift to Anthropic’s Claude chatbot.
Market data cited by TechCrunch from analytics firm Sensor Tower indicated a sharp surge in ChatGPT uninstallations shortly after the Pentagon deal was announced. At the same time, downloads of Claude increased significantly, helping the app climb to the top of the free app rankings in several countries.
Anthropic has also introduced new features designed to attract users from competing platforms, including tools that allow people to import conversation memories from ChatGPT and other chatbots. The company has expanded its memory capabilities to free users as well, a move widely seen as an attempt to capitalize on the backlash surrounding OpenAI’s government partnership.
