OpenAI recently announced a significant contract with the Pentagon, claiming it has secured an agreement that upholds strong prohibitions against domestic surveillance and the use of artificial intelligence in lethal military actions. CEO Sam Altman highlighted these commitments in a post on X (formerly Twitter) on February 27, stating, “Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems.” However, skepticism remains as the specifics of the contract are not publicly available.
This deal follows the collapse of negotiations between the U.S. Military and Anthropic, a leading competitor to OpenAI. Anthropic’s contract talks fell apart over its insistence on embedding similar prohibitions against killer robots and domestic spying, leading to a directive from then-President Donald Trump to phase out the use of Anthropic’s tools within six months. Given this backdrop, questions arise about how OpenAI could take on a contract without encountering the same issues.
OpenAI has attempted to clarify its position through several social media posts from executives, including Altman and Katrina Mulligan, the company’s national security chief. Altman has claimed that the firm negotiated stricter terms regarding domestic surveillance, yet the contract’s text has not been released to verify these assertions. The Department of Defense has not responded to requests for comment regarding the agreement.
Transparency Concerns
Despite Altman’s assurances, the lack of transparency surrounding the contract raises significant concerns. OpenAI has only shared snippets of the agreement’s language, filled with public relations jargon, leaving many to question the validity of their claims. Altman essentially asks the public to “trust” him, Trump, and Defense Secretary Pete Hegseth to uphold these commitments.
In a response to criticism, Altman stated that the company has worked with the Department of War (DoW) to clarify its principles. He mentioned that the Pentagon affirmed that OpenAI’s services would not be utilized by intelligence agencies, like the NSA, without a contract modification. However, the absence of the actual contract leaves this affirmation unverified.
Mulligan initially promised a detailed explanation of the contract terms but later declined to share specific language, citing no obligation to do so. She added that she would be open to collaborating with the NSA if appropriate safeguards were established, but failed to specify what those safeguards would entail.
Expert Opinions on OpenAI’s Contract
Former military officials have expressed grave concerns about the arrangement. Brad Carson, a former under secretary of the Army, stated, “I’m not confident in the language at all. And in some parts, I don’t even believe it.” He emphasized that blocking access for agencies like the NSA would limit OpenAI’s tools in critical intelligence contexts, such as ongoing military operations.
Another former Pentagon official, who requested anonymity, highlighted the ambiguous language surrounding “intentional” surveillance as a potential loophole. “That’s the get out of jail free card right there,” they said, indicating that the wording could allow for broader interpretations that might permit surveillance without explicit consent.
Calls for Disclosure
Alan Rozenshtein, a former attorney in the Department of Justice’s National Security Division, stated, “There is nothing OpenAI can do to clarify this except release the contract.” He criticized the lack of public access to the contract as “not sustainable” and “bizarre.” If OpenAI truly intends to restrict its tools from agencies known for extensive domestic surveillance, this commitment should be clearly documented in the contract.
Concerns have also been raised about misleading statements made by OpenAI officials. In response to a query about the contract allowing the Pentagon to analyze commercially available data, Mulligan claimed that the Pentagon had no legal authority to do so, a statement contradicted by a declassified 2022 report from the Office of the Director of National Intelligence detailing the government’s capacity to collect such data.
Implications and Future Directions
The implications of this contract extend beyond OpenAI and the Pentagon. The arrangement could set a precedent for future collaborations between tech companies and government agencies, raising questions about privacy, surveillance, and ethical responsibilities in an age of advanced artificial intelligence. The ongoing discussions around national security and technology integration will be pivotal as society navigates the balance between innovation and civil liberties.
As the situation develops, it is essential for the public, lawmakers, and oversight bodies to demand clarity and accountability from OpenAI and the Pentagon. The call for transparency is not just about this specific contract but reflects broader societal concerns regarding the intersection of technology and government power.
Moving forward, stakeholders will be watching closely for any further disclosures regarding the contract and the operational realities of how OpenAI’s technology will be employed. Engaging in this dialogue will be crucial as we collectively determine the ethical frameworks guiding the use of AI within national security.
We encourage readers to share their thoughts on this evolving story and the implications of AI in government contracts.
