OpenAI Revises Pentagon Agreement After Sam Altman Acknowledges Its ‘Careless’ Appearance

OpenAI is revisiting its recent agreement to provide artificial intelligence technologies to the U.S. Department of War (DoW) after its CEO expressed that the deal came off as “opportunistic and sloppy.” This reassessment follows significant backlash regarding the potential misuse of its AI for domestic mass surveillance.

In a statement made on Monday, OpenAI’s CEO, Sam Altman, acknowledged that the company would implement explicit restrictions to prevent its technology from enabling mass surveillance or being utilized by defense intelligence bodies such as the National Security Agency (NSA).

OpenAI, which boasts a user base of over 900 million for its ChatGPT platform, quickly initiated this contract after the Pentagon terminated its agreement with the previous AI contractor, Anthropic.

Anthropic had previously stated that “using these systems for mass domestic surveillance is incompatible with democratic values,” which led President Donald Trump to publicly criticize the company, labeling it as “leftwing nut jobs,” and ordering a halt to the federal government’s use of its technology.

Despite OpenAI’s assurances that the agreement did not permit surveillance applications, many commentators evoked the infamous Snowden scandal from 2013, which revealed the NSA’s large-scale collection of phone and internet data.

The agreement incited a wave of dissatisfaction among OpenAI users, with many on platforms like X and Reddit launching a campaign to “delete ChatGPT.” One user provocatively remarked, “You’re now training a war machine. Let’s see proof of cancellation.”

In the wake of this controversy, the AI chatbot Claude, developed by Anthropic, surged to the top of Apple’s App Store rankings, outperforming ChatGPT, as analyzed by Sensor Tower.

Altman addressed the situation in a message to employees on X, admitting that the deal, initially announced on a Friday, was put together too hastily following Anthropic’s dismissal.

“We shouldn’t have rushed to get this out on Friday,” Altman remarked. “These issues are incredibly complex and need clear communication. Our intention was to de-escalate a potential crisis, but the result was that it simply looked opportunistic and sloppy.”

When the deal was first presented, OpenAI claimed it had established “more guardrails than any prior agreements for classified AI deployment, including those made with Anthropic.”

Nonetheless, the military’s engagement with AI has raised concerns, prompting around 900 employees from both OpenAI and Google to sign an open letter urging their management not to allow the DoW to use their technologies for surveillance or autonomous warfare.

In their letter, employees expressed that the government was intentionally trying to create division within companies by instilling fear that the competition would relent. The employees stated: “We hope our leaders will put aside their differences and stand together to continue to refuse the DoW’s current demands for permission to use our models for domestic mass surveillance and autonomous killing without human oversight.”

The letter was endorsed by 796 employees from Google and 98 individuals from OpenAI. In a blog post announcing the deal with the DoW, OpenAI emphasized that one of its non-negotiable principles was “no use of OpenAI technology to direct autonomous weapons systems.”

However, some observers, including Miles Brundage, who previously led policy research at OpenAI, have raised questions about how OpenAI secured a deal that alleviates the ethical concerns that Anthropic found unmanageable. He commented on X, “OpenAI employees’ default assumption here should unfortunately be that OpenAI caved + framed it as not caving, and screwed Anthropic while framing it as helping them.”

Brundage also expressed his willingness to face imprisonment rather than comply with an unconstitutional government order, stating, “We want to work through democratic processes. It should be the government making the crucial societal decisions. We seek to have a voice and a seat at the table to share our expertise and advocate for the principles of liberty.”

In the midst of this turmoil, three additional cabinet-level agencies—the Department of State, the Treasury, and Health and Human Services—have opted to discontinue their use of Anthropic’s AI products following the DoW’s classification of the company as a supply chain risk. Trump has mandated that all U.S. government entities phase out Anthropic’s technology due to Secretary of Defense Pete Hegseth’s judgment.