Just a day after securing its own Pentagon contract, OpenAI moved quickly to clarify how its artificial intelligence tools would be handled inside the U.S. military’s classified systems.
The company said its agreement with the Defense Department includes stricter safeguards than any previous classified AI deployment — including arrangements previously made with Anthropic. According to OpenAI, the deal is designed to ensure that its technology is used responsibly, even within sensitive national security environments.
The announcement followed a dramatic shift in Washington. President Donald Trump ordered federal agencies to halt cooperation with Anthropic, and the Pentagon indicated it would label the startup a supply-chain risk. Such a designation could significantly limit the company’s ability to do business with the government. Anthropic has said it plans to challenge any such move in court.
Amid that fallout, OpenAI — backed by major investors including Microsoft, Amazon and SoftBank — confirmed that its own contract with the Pentagon enforces three firm boundaries. Its AI systems cannot be used for mass domestic surveillance, to operate autonomous weapons without human control, or to make high-stakes automated decisions independently.
The company described its protections as multi-layered. It said it retains full authority over its safety systems, deploys its models through secure cloud infrastructure, ensures cleared OpenAI personnel remain involved in operations, and relies on strong contractual clauses to reinforce those limits.
Over the past year, the Pentagon has signed agreements worth up to $200 million each with leading AI firms, including OpenAI, Anthropic and Google, as it pushes to integrate advanced software into defense planning. At the same time, military officials have sought broad operational flexibility, resisting efforts by tech companies to tightly restrict how AI tools may be applied in weapons or battlefield contexts.
OpenAI acknowledged that any violation of its contract terms by the government could result in termination of the agreement, though the company said it does not anticipate that outcome.
In a notable gesture, OpenAI also urged officials not to classify Anthropic as a supply-chain threat, saying it had already communicated its position to the government. The statement underscored the delicate balance now emerging between competition in Silicon Valley and the shared responsibility many tech leaders feel over how AI is used in matters of war and national security.
Also Read: OpenAI and Reliance Partner to Introduce AI Search on JioHotstar
