Anthropic cannot manipulate its generative AI model Claude once the US military has it running, an executive wrote in a court filing on Friday. The statement was made in response to accusations from the Trump administration about the company potentially tampering with its AI tools during war.

“Anthropic has never had the ability to cause Claude to stop working, alter its functionality, shut off access, or otherwise influence or imperil military operations,” Thiyagu Ramasamy, Anthropic’s head of public sector, wrote. “Anthropic does not have the access required to disable the technology or alter the model’s behavior before or during ongoing operations.”

The Pentagon has been sparring with the leading AI lab for months over how its technology can be used for national security—and what the limits on that usage should be. This month, defense secretary Pete Hegseth labeled Anthropic a supply-chain risk, a designation that will prevent the Department of Defense from using the company’s software, including through contractors, over the coming months. Other federal agencies are also abandoning Claude.

Anthropic filed two lawsuits challenging the constitutionality of the ban and is seeking an emergency order to reverse it. However, customers have already begun canceling deals. A hearing in one of the cases is scheduled for March 24 in federal district court in San Francisco. The judge could decide on a temporary reversal soon after.

In a filing earlier this week, government attorneys wrote that the Department of Defense “is not required to tolerate the risk that critical military systems will be jeopardized at pivotal moments for national defense and active military operations.”

The Pentagon has been using Claude to analyze data, write memos, and help generate battle plans, WIRED reported. The government’s argument is that Anthropic could disrupt active military operations by turning off access to Claude or pushing harmful updates if the company disapproves of certain uses.

Ramasamy rejected that possibility. “Anthropic does not maintain any back door or remote ‘kill switch,’” he wrote. “Anthropic personnel cannot, for example, log into a DoW system to modify or disable the models during an operation; the technology simply does not function that way.”

He went on to say that Anthropic would be able to provide updates only with the approval of the government and its cloud provider, in this case Amazon Web Services, though he didn’t specify it by name. Ramasamy added that Anthropic cannot access the prompts or other data military users enter into Claude.

Anthropic executives maintain in court filings that the company does not want veto power over military tactical decisions. Sarah Heck, head of policy, wrote in a court filing on Friday that Anthropic was willing to guarantee as much in a contract proposed March 4. “For the avoidance of doubt, [Anthropic] understands that this license does not grant or confer any right to control or veto lawful Department of War operational decision‑making,” the proposal stated, according to the filing, which referred to an alternative name for the Pentagon.

The company was also ready to accept language that would address its concerns about Claude being used to help carry out deadly strikes without human supervision, Heck claimed. But negotiations ultimately broke down.

For the time being, the Defense Department has said in court filings that it “is taking additional measures to mitigate the supply chain risk” posed by the company by “working with third-party cloud service providers to ensure Anthropic leadership cannot make unilateral changes” to the Claude systems currently in place.

United States congressman Jim Himes, the ranking Democrat on the House Intelligence Committee, is privately lobbying colleagues to preserve the FBI’s power to conduct warrantless searches of Americans’ communications, WIRED has learned, arguing that he has seen no evidence that the Trump administration is abusing its authority.

In a letter obtained by WIRED, Himes urges fellow Democrats to support the White House’s request to renew a controversial surveillance program that intercepts the electronic data of foreigners abroad. While targeted at foreigners, the program—authorized under Section 702 of the Foreign Intelligence Surveillance Act—also sweeps in vast quantities of private messages belonging to US citizens.

Himes’ pitch relies on the “56 reforms” passed by Congress in 2024, which codified the FBI’s own internal protocols as a substitute for constitutional warrants. In the letter, Himes claims these changes are “working as intended” to prevent domestic misuse, citing a compliance rate “exceeding 99 percent” over the past two years.

The structural foundations of that defense, however, have been fundamentally altered by recent changes within the FBI. Himes’ “99 percent” compliance metric was produced by the Office of Internal Auditing, for instance—a unit that long served as a smoke alarm designed to detect illegality, but no longer exists.

The unit was shuttered by FBI director Kash Patel last year. Historic court opinions based on its data had previously exposed hundreds of thousands of improper FBI searches. Without the auditors required to calculate failure rates, the compliance mechanisms Himes points to have effectively ceased to function.

In a statement, Himes’ office largely reiterated the positions laid out in his letter to colleagues. “I am open to making further reforms to Section 702, building on the many successful reforms we made in reauthorization legislation two years ago,” he says. “A short-term reauthorization of Section 702 will enable Congress to thoroughly debate the pros and cons of these suggested reforms—and to determine if compromise is possible—without placing our national security in peril by allowing the program to expire.”

As a member of the so-called Gang of Eight—a bipartisan group of lawmakers who are briefed on highly sensitive classified information—Himes possesses some of the deepest knowledge of the spy program. Nevertheless, his letter contains several other claims that appear fundamentally at odds with the mechanics of FISA oversight.

“Because of how heavily it is overseen by all three branches of government,” Himes says, “any effort to misuse the program would almost certainly become known to the Foreign Intelligence Surveillance Court and to Congress.”

The Foreign Intelligence Surveillance Court is a secret court that possesses no investigative arm to audit FBI databases. Similar to Congress, its oversight role is purely reactive, relying entirely on the US Justice Department to self-report violations.

“Neither Congress nor the FISA Court conducts independent audits of the FBI’s queries,” says Liza Goitein, senior director of the Brennan Center’s Liberty and National Security Program. “They rely on the Department of Justice to conduct thorough audits and to report the results truthfully and promptly. This particular Department of Justice has gutted internal oversight mechanisms and has been rebuked by dozens of federal courts for providing inaccurate, misleading, or incomplete information.”

There are no judges standing between the FBI and the private communications of millions of Americans, something that Himes and other members of his committee claim is necessary for the government to react quickly to terrorist threats. Critics argue that, given the current administration’s efforts to dismantle internal checks at the FBI, this is a massive vulnerability, leaving Americans exposed to surveillance abuses that will take years to declassify—if they’re ever reported at all.

After a day of widespread, overwhelming pushback, Nvidia CEO Jensen Huang doubled down and said gamers are “completely wrong” about DLSS. (You know how much gamers love being told that they’re wrong.) But developers at Capcom and Ubisoft say they didn’t even know what the tech demo would look like and, according to Insider Gaming, found out about it the same time everyone else did and were just as surprised. (Nvidia, Ubisoft, and Capcom did not immediately get back to our request for comment.)

“I think the reaction from gamers is understandable,” Marwan Mahmoud, a game developer at Incrypt, wrote in an email to WIRED. “Some games started relying too heavily on these technologies instead of focusing on proper optimization. From a developer perspective, it feels a bit different because you see DLSS as a tool that helps rather than a core solution.”

The problem for many people, developers included, is the one-size-fits-all approach of a technology that can adjust visuals across various game types.

“The artist has a style, the artist has an art direction that you’re going to give him, and that’s something that AI kind of doesn’t respect all the time,” says Raúl Izquierdo, an indie game developer in Mexico, “Maybe I don’t want my characters to be yassified.”

Bates agrees, saying he doesn’t think every game needs to be photo real. And that sentiment is also echoed by game developer Sterling Reames, who has worked at Striking Distance Studios and Zynga. “People just want better games,” Reames wrote in a message to WIRED. “That’s as plainly as I can put it.

At GTC, Nvidia ran its demo on its most powerful consumer graphics cards, two GeForce RTX 5090s. Had Nvidia made its selling point for the tech that it saves resources, thus enabling older hardware to deliver more impressive graphics, there may have been something to that.

“What’s the point if you’re not going to do it on weaker hardware?” Izquierdo says. “If this were done on an [RTX 2080 graphics card], for instance, I think I would be thinking differently about it. OK, this is for the betterment of gamers’ experiences and everything, not just for selling graphic cards.”

Ultimately, Nvidia’s demo, and GTC writ large, was a flex of the company’s power in the AI space. The reaction, Bates posits, is more about humans dealing with not just crossing the uncanny valley, but what happens when we reach the other side.

“Right now it’s pretty clearly a thing they are forced to do to demonstrate their prowess as an AI company,” Bates says. “But the truth is, this is going to be the default in a few years, and nobody is even going to think twice about it. It’s Jensen’s world, we’re just living in it.”