Sam Altman’s Defense of OpenAI’s Pentagon Contract
OpenAI’s CEO, Sam Altman, publicly defended the company’s new contract with the Pentagon over the weekend, coinciding with President Trump’s directive for federal agencies to sever ties with competitor Anthropic.
Following a U.S.-Israel offensive against Iran, Altman discussed the partnership allowing the Department of the Army to utilize OpenAI’s artificial intelligence models in sensitive networks. “I want to clarify our work with the Department of the Army and share our thoughts from the past few days,” he mentioned.
He emphasized that security and equitable benefit-sharing from AI are central to OpenAI’s mission. “We hold two critical principles: no domestic mass surveillance and clear accountability for the use of force, particularly concerning autonomous weapons systems,” he stated. The Department of the Army aligns with these principles, which have been integrated into their agreement.
OpenAI’s Position Amidst Military AI Controversy
This new agreement was announced shortly after Trump ordered federal agencies to stop using Anthropic’s technology, designating them as a “supply chain risk to national security.” Secretary of the Army Pete Hegseth stated this publicly, stirring discussions on military AI usage.
Anthropic’s CEO, Dario Amodei, had previously declined the Army’s request for permission to use their technology for “all lawful purposes,” citing concerns about surveillance and autonomous weaponry.
In discussing why OpenAI was accepted while Anthropic was not, Altman noted, “Anthropic seemed to focus more on specific prohibitions in their contract rather than applicable laws, and we were fine with our approach.” He hinted that Anthropic might have sought more operational control than what OpenAI proposed.
Implications of the Agreement
Altman stated that the Army made no threats prior to the contract and expressed surprise regarding OpenAI’s willingness to consider declassification. Originally, they intended to work only on unclassified projects with the Department. However, earlier discussions led to a quicker engagement regarding classified work.
“We believe the Department needs a reliable AI partner, even if working in classified settings is more complicated. We had previously turned down deals similar to what Anthropic accepted,” he elaborated, highlighting flexibility on the Army’s part.
Altman also addressed concerns about the hastiness of the agreement, suggesting they aimed to “de-escalate” tensions surrounding the situation. He mentioned, “The current trajectory raises risks for healthy competition and innovation in America.”
Despite the partnership, Altman expressed worries about potential legal incidents exposing OpenAI to the same risks as Anthropic. “If we must face that, we will, but it carries some risk for us,” he acknowledged. His motivation for acting swiftly was to possibly improve the situation’s outcome.
Future of AI Development and Oversight
When asked whether the government might consider nationalizing OpenAI, Altman seemed uncertain, sharing, “I think about it, sure, but it seems quite unlikely right now.” He did, however, stress the importance of a close relationship between government entities and technology developers.
Altman touched on the challenging aspects of the agreement, particularly concerning “out-of-home surveillance.” While he understands the military’s need for surveillance on foreign nationals, he expressed discomfort with the implications this has. “It’s crucial for society to contemplate the ramifications of this technology,” he said, stressing the importance of democratizing AI development despite the complexities of military oversight.
In closing, Altman pointed out the importance of dialogue and consideration regarding policy implications on technology, indicating that while there’s concern, it’s a process that belongs to society as a whole.
