The rapidly escalating conflict between Anthropic and the Pentagon, which started when the company refused to let the government use its technology to spy on Americans, has now gone to court. The Department of Defense retaliated by designating the company a “supply chain risk” (SCR). Now, Anthropic is asking courts to block the designation, arguing that the First Amendment does not permit the government to coerce a private actor to rewrite its code to serve government ends.

We agree.

As EFF, the Foundation for Individual Rights and Expression, and multiple other public interest organizations explained in a brief filed in support of Anthropic’s motion, the development and operation of large language models involve multiple expressive choices protected by the First Amendment. Requiring a company to rewrite its code to remove guardrails means compelling different expression, a clear constitutional violation. Further, the public record shows that the SCR designation is intended to punish the company both for pushing back and for its CEO’s public statements explaining that AI may supercharge surveillance practices that current law has proven ill-equipped to address.

As we also explain, the company’s concerns about how the government will use its technology are well-founded. The U.S. government has a long history of illegally surveilling its citizens without adequate judicial oversight based on questionable interpretations of its Constitutional and statutory obligations. The Department of Defense acquires vast troves of personal information from commercial entities, including individuals’ physical location, social media, and web browsing data. Other government agencies continue to collect and query vast quantities of Americans’ information, including by acquiring information from third party data brokers.

A growing body of social science research illustrates the chilling effects of these pervasive activities. Fearing retribution for unpopular views, dissenters stay silent. And AI only exacerbates the problem. AI can quickly analyze the government’s massive datasets or combine that information with data scraped off the internet, purchased through the commercial data broker market, or from local police surveillance devices and use all of that data to construct a comprehensive picture of a person’s life and infer sensitive details like their religious beliefs, medical conditions, political opinions, or even sex partners. For example, an agency could use AI to infer an individual’s association with a particular mosque based on data showing that they visited its website, followed its social media accounts, and were located near the mosque during religious services. AI can also deanonymize online speech by using public information to unmask anonymous users.

It is easy to conceive how an agency, a government employee with improper intent, or a malicious hacker could exploit these capabilities to monitor public discourse, preemptively squelch dissent, or persecute people from marginalized communities. Against this background and absent meaningful changes to the governing national security laws and judicial oversight structure, it is entirely reasonable for Anthropic—or any other company—to insist on its own guardrails.

Without action from Congress, the task of protecting your privacy has fallen in large part to Big Tech—something no one wants, including Big Tech. But if Congress won’t do it, companies like Anthropic must be allowed to step in, without facing retribution.

OpenAI, the maker of ChaptGPT, is rightfully facing widespread criticism for its decisions to fill the gap the U.S. Department of Defense (DoD) created when rival Anthropic refused to drop its restrictions against using its AI for surveillance and autonomous weapons systems. After protests from both users and employees who did not sign up to support government mass surveillance—early reports show that ChaptGPT uninstalls rose nearly 300% after the company announced the deal—Sam Altman, CEO of OpenAI, conceded that the initial agreement was “opportunistic and sloppy.” He then re-published an internal memo on social media stating that additions to the agreement made clear that “Consistent with applicable laws, including the Fourth Amendment to the United States Constitution, National Security Act of 1947, [and] FISA Act of 1978, the AI system shall not be intentionally used for domestic surveillance of U.S. persons and nationals.”

Trouble is, the U.S. government doesn’t believe “consistent with applicable laws” means “no domestic surveillance.” Instead, for the most part, the government has embraced a lax interpretation of “applicable law” that has blessed mass surveillance and large-scale violations of our civil liberties, and then fought tooth and nail to prevent courts from weighing in. 

"After all, many of the world’s most notorious human rights atrocities have historically been “legal” under existing laws at the time."

“Intentionally” is also doing an awful lot of work in that sentence. For years the government has insisted that the mass surveillance of U.S. persons only happens incidentally (read: not intentionally) because their communications with people both inside the United States and overseas are swept up in surveillance programs supposedly designed to only collect communications outside the United States. 

The company’s amendment to the contract continues in a similar vein, “For the avoidance of doubt, the Department understands this limitation to prohibit deliberate tracking, surveillance, or monitoring of U.S. persons or nationals, including through the procurement or use of commercially acquired personal or identifiable information.” Here, “deliberate” is the red flag given how often intelligence and law enforcement agencies rely on incidental or commercially purchased data to sidestep stronger privacy protections.

Here’s another one: “The AI System shall not be used for unconstrained monitoring of U.S. persons’ private information as consistent with these authorities. The system shall also not be used for domestic law-enforcement activities except as permitted by the Posse Comitatus Act and other applicable law.” What, one wonders, does “unconstrained” mean, precisely—and according to whom? 

Lawyers sometimes call these “weasel words” because they create ambiguity that protects one side or another from real accountability for contract violations. As with the Anthropic negotiations, where the Pentagon reportedly agreed to adhere to Anthropic’s red lines only “as appropriate,” the government is likely attempting to publicly commit to limits in principle, but retain broad flexibility in practice.

OpenAI also notes that the Pentagon promised the NSA would not be allowed to use OpenAI’s tools absent a new agreement, and that its deployment architecture will help it verify that no red lines are crossed. But secret agreements and technical assurances have never been enough to rein in surveillance agencies, and they are no substitute for strong, enforceable legal limits and transparency.

OpenAI executives may indeed be trying, as claimed, to use the company’s contractual relationship with the Pentagon to help ensure that the government should use AI tools only in a way consistent with democratic processes. But based on what we know so far, that hope seems very naïve.

Moreover, that naïvete is dangerous. In a time when governments are willing to embrace extreme and unfounded interpretations of “applicable laws,” companies need to put some actual muscle behind standing by their commitments. After all, many of the world’s most notorious human rights atrocities have historically been “legal” under existing laws at the time. OpenAI promises the public that it will  “avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power,” but we know that enabling mass surveillance does both.     

OpenAI isn’t the only consumer-facing company that is, on the one hand, seeking to reassure the public that they aren’t participating in actions that violate human rights while, on the other, seeking to cash in on government mass surveillance efforts.  Despite this marketing double-speak, it is very clear that companies just cannot do both. It’s also clear that companies shouldn’t be given that much power over the limits of our privacy to begin with. The public should not have to rely on a small group of people—whether CEOs or Pentagon officials—to protect our civil liberties.

The U.S. military has officially ended its $200 million contract with AI company Anthropic and has ordered all other military contractors to cease use of their products. Why? Because of a dispute over what the government could and could not use Anthropic’s technology to do. Anthropic had made it clear since it first signed the contract with the Pentagon in 2025 that it did not want its technology to be used for mass surveillance of people in the United States or for fully autonomous weapons systems. Starting in January, that became a problem for the Department of Defense, which ordered Anthropic to give them unrestricted use of the technology. Anthropic refused, and the DoD retaliated.

There is a lot we could learn from this conflict, but the biggest take away is this: the state of your privacy is being decided by contract negotiations between giant tech companies and the U.S. government—two entities with spotty track records for caring about your civil liberties. It’s good when CEOs step up and do the right thing—but it's not a sustainable or reliable solution to build our rights on. Given the government’s loose interpretations of the law, ability to find loopholes to surveil you, and willingness to do illegal spying, we needs serious and proactive legal restrictions to prevent it from gobbling up all the personally data it can acquire and using even routine bureaucratic data for punitive ends.

Imposing and enforcing such those restrictions is properly a role for Congress and the courts, not the private sector. 

The companies know this. When speaking about the specific risk that AI poses to privacy, the CEO of Anthropic Dario Amodei said in an interview, “I actually do believe it is Congress’s job. If, for example, there are possibilities with domestic mass surveillance—the government buying of bulk data has been produced on Americans, locations, personal information, political affiliations, to build profiles, and it’s not possible to analyze all of that with AI—the fact that that is legal—that seems like the judicial interpretation of the Fourth Amendment has not caught up or the laws passed by Congress have not caught up.” 

The example he cites here is a scarily realistic one—because it’s already happening. Customs and Border Protection has tapped into the online advertising world to buy data on Americans for surveillance purposes. Immigration and Customs Enforcement has been using a tool that maps millions of peoples’ devices based on purchased cell phone data. The Office of the Director of National Intelligence has proposed a centralized data broker marketplace to make it easier for intelligence agencies to buy commercially available data. Considering the government’s massive contracts with a bunch of companies that could do analysis, including Palantir, a company which does AI-enabled analysis of huge amounts of data, then the concerns are incredibly well founded. 

But Congress is sadly neglecting its duties. For example, a bill that would close the loophole of the government buying personal information passed the House of Representatives in 2024, but the Senate stopped it.  And because Congress did not act, Americans must rely on a tech company CEO has to try to protect our privacy—or at least refuse to help the government violate it.

Privacy in the digital age should be an easy bipartisan issue. Given that it’s wildly popular (71% of American adults are concerned about the government's use of their data and among adults that have heard of AI 70% have little to no trust in how companies use those products) you would think politicians would be leaping over each other to create the best legislation and companies would be promising us the most high-end privacy protecting features. Instead, for the time being, we are largely left adrift in a sea of constant surveillance, having to paddle our own life rafts.

EFF has, and always will, fight for real and sustainable protections for our civil liberties including  a world where our privacy does not rest upon the whims of CEOs and back room deals with the surveillance state. 

The Secretary of Defense has given an ultimatum to the artificial intelligence company Anthropic in an attempt to bully them into making their technology available to the U.S. military without any restrictions for their use. Anthropic should stick by their principles and refuse to allow their technology to be used in the two ways they have publicly stated they would not support: autonomous weapons systems and surveillance. The Department of Defense has reportedly threatened to label Anthropic a “supply chain risk,” in retribution for not lifting restrictions on how their technology is used. According to WIRED, that label would be, “a scarlet letter usually reserved for companies that do business with countries scrutinized by federal agencies, like China, which means the Pentagon would not do business with firms using Anthropic’s AI in their defense work.”

Anthropic should stick by their principles and refuse to allow their technology to be used in the two ways they have publicly stated they would not support: autonomous weapons systems and surveillance.

In 2025, reportedly Anthropic became the first AI company cleared for use in relation to classified operations and to handle classified information. This current controversy, however, began in January 2026 when, through a partnership with defense contractor Palantir, Anthropic came to suspect their AI had been used during the January 3 attack on Venezuela. In January 2026, Anthropic CEO Dario Amodei wrote to reiterate that surveillance against US persons and autonomous weapons systems were two “bright red lines” not to be crossed, or at least topics that needed to be handled with “extreme care and scrutiny combined with guardrails to prevent abuses.” You can also read Anthropic’s self-proclaimed core views on AI safety here, as well as their LLM, Claude’s, constitution here. 

Now, the U.S. government is threatening to terminate the government’s contract with the company if it doesn’t switch gears and voluntarily jump right across those lines.  

Companies, especially technology companies, often fail to live up to their public statements and internal policies related to human rights and civil liberties for all sorts of reasons, including profit. Government pressure shouldn’t be one of those reasons. 

Whatever the U.S. government does to threaten Anthropic, the AI company should know that their corporate customers, the public, and the engineers who make their products are expecting them not to cave. They, and all other technology companies, would do best to refuse to become yet another tool of surveillance.

We recently introduced a policy governing large language model (LLM) assisted contributions to EFF's open-source projects. At EFF, we strive to produce high quality software tools, rather than simply generating more lines of code in less time. We now explicitly require that contributors understand the code they submit to us and that comments and documentation be authored by a human.

LLMs excel at producing code that looks mostly human generated, but can often have underlying bugs that can be replicated at scale. This makes LLM-generated code exhausting to review, especially with smaller, less resourced teams. LLMs make it easy for well-intentioned people to submit code that may suffer from hallucination, omission, exaggeration, or misrepresentation.

It is with this in mind that we introduce a new policy on submitting LLM-assisted contributions to our open-source projects. We want to ensure that our maintainers spend their time reviewing well thought out submissions. We do not completely outright ban LLMs, as their use has become so pervasive a blanket ban is impractical to enforce.

Banning a tool is against our general ethos, but this class of tools comes with an ecosystem of problems. This includes issues with code reviews turning into code refactors for our maintainers if the contributor doesn’t understand the code they submitted. Or the sheer scale of contributions that could come in as AI generated code but is only marginally useful or potentially unreviewable. By disclosing when you use LLM tools, you help us spend our time wisely.

EFF has described how extending copyright is an impractical solution to the problem of AI generated content, but it is worth mentioning that these tools raise privacy, censorship, ethical, and climatic concerns for many. These issues are largely a continuation of tech companies’ harmful practices that led us to this point. LLM generated code isn’t written on a clean slate, but born out of a climate of companies speedrunning their profits over people. We are once again in “just trust us” territory of Big Tech being obtuse about the power it wields. We are strong  advocates of using tools to innovate and come up with new ideas. However, we ask you to come to our projects knowing how to use them safely.