OpenAI is revising the Department of Defense’s deal after a setback

OpenAI CEO Sam Altman says the company rushed its latest deal with the US Department of Defense (DOW), admitting it appeared “opportunistic and sloppy.” In an internal memo shared with X, Altman said OpenAI is now amending its contract to provide the military with AI technology. It seems to have done little to ease the anxiety.

“[W]I shouldn’t have rushed this out on Friday,” Altman wrote in an X post on Monday. “The issues are very complex, and require clear communication. We were honestly trying to play things down and avoid the worst possible outcome, but I think it looks opportunistic and sloppy.”

OpenAI announced its partnership with DOW late last week, snapping up the contract within days of President Donald Trump ordering federal agencies to stop using rival Anthropic. According to Anthropic CEO Dario Amodei, the split was because it refused DOW’s demands to remove safeguards on the use of AI in mass surveillance at home and autonomous weapons. Instead, the DOW wanted to use Anthropic’s AI tools for “any lawful use.”

Therefore, OpenAI’s immediate DOW deal sparked a backlash from its regular users. Despite OpenAI’s claim that its agreement has more safeguards than the original Anthropic agreement, the contract appears to allow for mass surveillance and AI-controlled weapons as long as such use is legal, and conditions are placed on where it will be allowed.

Now OpenAI is trying to do damage control, saying it worked with the DOW to add new language to the contract specifically addressing the use of its technology in home surveillance.

“In all our discussions, the Department [of War] has made it clear that it shares our commitment to ensuring that our tools cannot be used for domestic surveillance,” OpenAI wrote Monday in an update to its initial announcement.

OpenAI is revising the Department of Defense’s deal after a setback

Unfortunately, the new amendments shared by OpenAI continue to rely on legality as a deterrent to mass surveillance, leaving such uses possible if the US government changes the law. They also failed to address the issue of autonomous weapons.

“Pursuant to applicable laws… the AI ​​system may not be used for domestic surveillance of US citizens and nationals,” the new sections read. “For the avoidance of doubt, the Department understands this limitation to prevent the intentional tracking, surveillance, or surveillance of US or international persons, including the purchase or use of commercially obtained or identifiable personal information.”

This Tweet is currently unavailable. Either it is loading or it has already been downloaded.

Many social media users reacted to the OpenAI contract changes with skepticism, with some saying that its “deliberate” prohibition on surveillance leaves significant loopholes.

“It’s hard not to read as an acknowledgment of the AI ​​dragnet,” political analyst Tyson Brody (@tysonbrody) responded to Altman’s post. “‘Deliberately’ and ‘deliberately’ – so the American people the will can be swept from this data, but the government can demand ‘risk accumulation’ and thus be legal.”

“‘Not intentionally used’ is not a real defense against an autonomous AI system,” wrote @Andy_Bloch. “It can end up being monitored because of what it was trained for, how it counts, or how people use it afterwards.”

Altman previously stated that OpenAI would limit the use of its AI tools to legal, not ethical, lines during a Q&A held shortly after the DOW deal was announced. The CEO expressed reluctance to take ethical action, saying that OpenAI prefers to follow government guidelines rather than consider these issues on its own.

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

“It should be government that makes important decisions about 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 freedom. But we are clear about how the system works (because many people have asked, if I received what I believed to be an unconstitutional order, of course I would rather go to jail than follow it).”

Altman said that DOW intelligence agencies such as the National Security Agency (NSA) will not use OpenAI technology without amending their contract. However, it currently seems unlikely that OpenAI would deny legal requests for such corrections, regardless of the ethical issues that may arise. (The NSA was previously revealed to have been spying on US citizens through whistleblower Edward Snowden in 2013.)

Many OpenAI customers canceled their ChatGPT subscriptions due to the company’s deal with the DOW, as issuance reportedly jumped 295 percent after the news. Anthropic’s AI chatbot Claude has since dethroned ChatGPT as the most downloaded free app in the US Apple App Store,

Disclosure: Ziff Davis, Mashable’s parent company, in April 2025 filed a lawsuit against OpenAI, alleging that it infringes Ziff Davis’s copyright in training and using its AI programs.