Sam Altman Defends OpenAI's Pentagon Deal, Criticizes 'Supply Chain Risk' Threat to Anthropic
Key Takeaways
- ▸OpenAI signed a rushed deal with the Pentagon partly to de-escalate tensions after Anthropic was threatened with a 'Supply-Chain Risk' designation
- ▸The agreement includes safeguards against domestic mass surveillance and requires human control over autonomous weapons, with similar terms to be offered to competing AI labs
- ▸Altman publicly criticized the Pentagon's threat against Anthropic, calling it dangerous for the industry despite OpenAI's competing interests
Summary
OpenAI CEO Sam Altman held an impromptu Q&A session on X.com to address the company's controversial deal with the U.S. Department of Defense (referred to as 'Department of War' in the original reporting). The partnership came after failed negotiations between the Pentagon and Anthropic, which resulted in threats to designate Anthropic as a 'Supply-Chain Risk to National Security.' Altman acknowledged the rushed nature of the deal and its poor optics, but defended it as a necessary de-escalation effort.
Altman revealed that OpenAI negotiated specific protections into the agreement, including prohibitions against domestic mass surveillance and requirements for 'human responsibility' in autonomous weapon systems. He emphasized that similar terms would be offered to other AI labs, positioning the deal as an industry-wide protective measure rather than a competitive advantage. The CEO also broke from his company's previous stance of doing 'non-classified work only,' citing the Department of Defense's flexibility on OpenAI's requirements.
In a notable departure from typical corporate positioning, Altman openly criticized the Pentagon's threat against Anthropic, calling it 'very bad for our industry and our country.' He expressed hope that the designation would be reversed, even if OpenAI faced backlash for the criticism. Altman framed the situation as symptomatic of tension between the AI industry and government, noting the contradictory messages companies send by warning about AI's geopolitical importance while refusing to assist defense efforts. He acknowledged personal concern about OpenAI itself potentially being deemed a supply chain risk if future disputes arise over legal interpretations of the agreement.
- OpenAI reversed its longstanding policy of doing only non-classified work with the government
- Altman acknowledged the deal exposes OpenAI to risk of similar supply chain designations if future disputes arise
