OpenAI updates Department of War deal after backlash

OpenAI CEO Sam Altman said the company rushed its recent deal with the US Department of War (DOW), admitting it seemed “opportunistic and sloppy”. In an internal memo he later shared on X, Altman said OpenAI was now modifying its agreement to provide the military with AI technology. This does not appear to have done much to allay concerns.

“[W]We shouldn’t have rushed to publish this on Friday,” Altman wrote in an X article on Monday. “The issues are extremely complex and require clear communication. We were genuinely trying to de-escalate the situation and avoid a much worse outcome, but I think it just seemed opportunistic and sloppy. »

OpenAI announced its partnership with the DOW late last week, landing the contract days after President Donald Trump ordered federal agencies to stop using its competitor Anthropic. According to Anthropic CEO Dario Amodei, the split is due to the DOW’s refusal to remove safeguards against the use of AI for mass domestic surveillance and fully autonomous weapons. Instead, the DOW wanted to use Anthropic’s AI tools for “any lawful use.”

As such, OpenAI’s early agreement with DOW prompted an immediate backlash from its civilian users. Despite OpenAI’s claim that its deal has even more safeguards than Anthropic’s original deal, the contract appears to allow both mass surveillance and AI-controlled weapons as long as such use is legal, and even defines the circumstances under which this would be permitted.

OpenAI is now trying to do damage control, saying it has worked with the DOW to add new language to the contract directly addressing the use of its technology for domestic surveillance.

“Throughout our discussions, the Ministry [of War] has made clear that it shares our commitment to ensuring that our tools will not be used for domestic surveillance,” OpenAI wrote Monday in an update to its initial announcement of the deal.

OpenAI updates War Department agreement after backlash

Unfortunately, the new amendments shared by OpenAI continue to rely on legality as a restrictive limit preventing mass surveillance, leaving such use possible if the US government changed the law. They also fail to resolve the issue of autonomous weapons.

“Consistent with applicable laws…the AI ​​system shall not be used intentionally for domestic surveillance of U.S. persons and nationals,” the new sections state. “For the avoidance of doubt, the Department understands that this limitation prohibits the deliberate tracking, surveillance, or control of persons or U.S. persons, including through the obtaining or use of commercially acquired personal or identifiable information.”

This Tweet is currently unavailable. It may be loading or has been deleted.

Many social media users have reacted with skepticism to OpenAI’s contract changes, with some saying the specific ban on “deliberate” surveillance leaves notable loopholes.

“It’s hard not to admit that this is a dragnet for AI,” political researcher Tyson Brody (@tysonbrody) responded to Altman’s post. “‘intentionally’ and ‘deliberately’ – so the Americans will “

“‘Not intentionally used’ is not a real protection in an autonomous AI system,” wrote @Andy_Bloch. “He may end up doing surveillance because of what he was trained on, what he found out, or how people use him afterward.”

Altman previously indicated that OpenAI would limit the use of its AI tools solely on legal, not ethical, grounds during a Q&A held shortly after the DOW deal was announced. The CEO has expressed reluctance to take an ethical stance, saying OpenAI prefers to follow government instructions rather than examine these issues itself.

Despite criticism of this apparent abdication of responsibility, Altman reiterated this position in his new memo, presenting it as deference to “democratic processes.”

“It should be the government that makes key decisions regarding society,” Altman wrote. “We want to have a voice and a seat at the table where we can share our expertise and fight for the principles of liberty. But we are clear about how the system works (because many people have asked me if I received what I believed to be an unconstitutional order, of course I would rather go to jail than follow it).”

Altman said DOW intelligence agencies such as the National Security Agency (NSA) would not use OpenAI technology without an amendment to their contract. Even so, it currently seems unlikely that OpenAI will refuse legal requests for such changes, regardless of any ethical issues that might arise. (The NSA was previously revealed to be conducting mass surveillance of US citizens by whistleblower Edward Snowden in 2013.)

Many OpenAI customers canceled their ChatGPT subscriptions in response to the company’s deal with the DOW, with uninstalls reportedly jumping 295% following the news. Claude, Anthropic’s AI chatbot, has since dethroned ChatGPT as the most downloaded free app on the US Apple App Store,

Disclosure: Ziff Davis, the parent company of Mashable, filed a lawsuit in April 2025 against OpenAI, alleging that it had violated Ziff Davis’ copyrights in the training and operation of its AI systems.