Openai has maintained several government-support initiatives, including a test partnership with the National Labs and Chatgpt Gov. Last week, company Announced It is rolling all of them under a single umbrella initiative: Openai for the government.
(Disclosure: ZDNET’s original company Ziff Davis filed a case of April 2025 against Openai, alleging that it violates Ziff Davis copyright training and operating its AI system.)
Pilot Program with DOD
The first priority of the initiative will be a pilot program with the US Department of Defense (DOD), which is overshadowed by $ 200 million, “for identifying and prototype on how Frontier AI (DOD) can change administrative operations, improving how the members of the service and their families get health care, how they see the program and acquisition data, support the proactive cyber defense”.
Also: How much energy does AI actually use? The answer is amazing – and slightly complex
The company said that all use cases in the contract should follow Openai policies and guidelines.
In April 2024, Microsoft allegedly picked Dall-E as a battlefield training tool for the Defense Department, arguing on Openai’s use policies. Since its installation, Openai’s Policies page It was said that its model is needed Should not be used for military developmentBut in January 2024, the company removed “military” and “war” from the use language.
The page now refuses to “harm yourself or others”, “to develop or use weapons, injure others or destroy property.”
By naming cases ranging from administration to cyber defense, the announcement holds a certain amount of flexibility – and, to some extent, ambiguity – how the technology of openi will be used in practice.
Too: AI agents will threaten humans to achieve their goals, get anthropic report
“DOD will be looking for devices that can be integrated into their complex and highly safe infrastructure,” said Ben Wan Van Ro, CEO and co-founder of Legion Intelligence, CEO and co-founder. “Openai’s new government offering can begin with some like chat villages, but real value will be realized when the model can be embedded in real applications and workflows, and can operate within various environments – including classified, humiliated and disconnected networks.”
Chatgpt Enterprise and access to the village
Openai for the government is not setting its places only on DOD.
As part of the new initiative, Openai Chatgpt is creating the “most capable model within safe and obedient environment” available to federal, state and local government workers via Enterprise and village. Those workers will also get the support dedicated by the preview of OpenAI, Custom Models for National Security, and the preview of upcoming openiI products and facilities, so that teams can plan to integrate them.
“Our goal is to unlock AI solutions that increase the capabilities of government employees, help them cut red tape and paperwork, and let them do more to come to work every day: serve American people,” reads the announcement.
Van Ruo noticed that openi’s move was part of a major marine change in the technical field. “Major AI organizations are rapidly entangled with DOD,” he said. “Recently, many have maintained their distance from federal space, but are now actively constructing it.”
Also: 15 new jobs can create AI – can anyone be your next gig?
In practice, this is partially because DOD has much money in its disposal than other government agencies. Own AI devices are not cheap, especially when they should be sewn to a hyper-system environment. “It needs to work within the inheritance systems in a variety of networks, and its guidelines and restrictions,” Van Ruo notes. He said that the most “forward-shout” among DOD agencies has also been when it comes to AI adoption.
President Trump’s AI Action Plan
More widely, however, this step is taking place against President Donald Trump’s background of administration’s approach to AI policy. The administration is ready to give its AI action plan by 22 July, a deadline trump set when he withdrew the former President Joe Biden’s executive order on AI in January.
Since assuming office, Trump has reduced the AI security guard within the government and cut the AI research funding – in the alarm of industry leaders – expanding partnership with AI companies, including Openi and Anthropic.
Also: The new AI models of anthropic for classified information are already in use by US village
Meanwhile, Trump’s so-called “Big, Beautiful Bill,” or HR1, the primary piece of his administration, includes a 10-year-old adjournment AI acceleration on the EON-state AI law in years, which will leave all regulation and enforcement at the federal level if passed if passed.
As the July deadline inch is close to the time limit, it seems that the Trump administration is outsourcing the wholesale of AI policy to contract with private AI companies rather than cultivating independent regulation.
Want more stories about AI? Sign up for innovationOur weekly newspapers.
