By law, on April 20, 2026, a broad tool for intercepting the communications of foreigners abroad – Section 702 of the Foreign Intelligence Surveillance Act – will sunset, unless Congress decides to renew it. This deadline gives Congress an unusual opportunity to examine the executive branch’s use of this surveillance law, which has been criticized for its ability to sweep in communications of Americans, and to consider its use in the context of other government surveillance tools and authorities.
To be sure, having worked in the government, we believe that historically Section 702 has provided the government with an indispensable national security tool with respect to the NSA and other agencies’ detecting foreign threats. But there has also been meaningful bipartisan interest in reforms to the tool. We recognize important long-standing questions and concerns it raises about Americans’ and others’ privacy rights and the parameters governing how the law functions.
What makes this moment different is the convergence of two factors: (1) the government’s rapidly expanding deployment of artificial intelligence in its surveillance programs; and (2) greater distrust of government actions involving Americans’ First Amendment political freedoms and Fourth Amendment privacy rights, which courts have repeatedly found to have been violated by this administration.
In order to facilitate robust deliberation, we provide the following questions for Members of Congress to ask executive branch officials as part of the Section 702 renewal process. Journalists and the public could – and should – also demand answers to these questions.
On Claude and Defense Department Surveillance
Anthropic, the company that has developed the Claude suite of AI tools, has sued the Department of Defense (DOD). In its complaint, the company objects to removing constraints on Claude’s use for “mass surveillance of Americans.” In its successful motion for a preliminary injunction, Anthropic objects, among other things, to “allowing Claude’s use on classified systems to surveil Americans en masse.”
The questions for Members of Congress to ask are:
- Under what specific circumstances does the U.S. government believe it can engage in mass surveillance of Americans? For what purposes? Relying on what authorities? With what prior authorization and oversight? Using which tools?
- How does the administration seek to use Claude (or other AI) to surveil Americans?
- Is any such program approved by a court, including the FISC (the Foreign Intelligence Surveillance Court)? Has the FISA Court been given a comprehensive description of all AI systems currently operating in the Section 702 program? In other surveillance programs?
- Under what circumstances specifically do you believe the DOD could engage in surveillance of Americans? What potential use cases for AI does it believe would be lawful in such circumstances, and why?
- Given the very narrow circumstances in which the U.S. military may ever be used for domestic purposes, why does the DOD have any need or intent to surveil Americans? Would such use depend on invocation of the Insurrection Act?
- What can be made public about any such program, so we do not have a recurrence of the “215” program (a surveillance program involving millions of Americans’ telephone metadata, which was considered unwise if not illegal by many after being revealed by Edward Snowden)?
- Do any other executive departments or agencies have any initiatives of any sort, whether planned or existing, for using AI for the mass surveillance of U.S. persons? If so, which AI companies’ products would be used for those purposes? If contracts or subcontracts exist, please provide those contracts and subcontracts.
On the Fourth Amendment
The Department’s use or intended use of AI may run afoul of the Fourth Amendment.
The constitutional framework: In Carpenter v. United States (2018), the Supreme Court held that the FBI’s warrantless acquisition of an individual’s cell site location information over several months violated the Fourth Amendment. The Court wrote: “Although such records are generated for commercial purposes, that distinction does not negate Carpenter’s anticipation of privacy … [T]he timestamped data provides an intimate window into a person’s life, revealing not only his particular movements, but through them his ‘familial, political, professional, religious, and sexual associations.’” The Court was quoting Justice Sotomayor’s concurrence in United States v. Jones (2012), in which she wrote: “GPS monitoring generates a precise, comprehensive record of a person’s public movements that reflects a wealth of detail about her familial, political, professional, religious, and sexual associations.”
The factual record: Recent statements by Anthropic executives about their objections to DOD’s use of Claude for mass surveillance of Americans appear to describe similar facts.
A Feb. 26, 2026 statement by Anthropic CEO Dario Amodei said:
“Mass domestic surveillance. We support the use of AI for lawful foreign intelligence and counterintelligence missions. But using these systems for mass domestic surveillance is incompatible with democratic values. AI-driven mass surveillance presents serious, novel risks to our fundamental liberties. To the extent that such surveillance is currently legal, this is only because the law has not yet caught up with the rapidly growing capabilities of AI. For example, under current law, the government can purchase detailed records of Americans’ movements, web browsing, and associations from public sources without obtaining a warrant, a practice the Intelligence Community has acknowledged raises privacy concerns and that has generated bipartisan opposition in Congress. Powerful AI makes it possible to assemble this scattered, individually innocuous data into a comprehensive picture of any person’s life—automatically and at massive scale.” (emphasis added)
In a declaration in a court filing, co-founder of Anthropic and Chief Science Officer, Jared Kaplan stated:
“Tools like Claude enable aggregation and analysis of massive datasets at unprecedented scale, potentially facilitating practices inconsistent with Americans’ rights even if they appear arguably compliant with laws written before the advent of AI and interpreted by courts only in a pre-AI context. For example, while there is generally no expectation of privacy in public spaces, powerful AI could enable the government to aggregate and analyze millions of public surveillance camera feeds into real-time, population-scale tracking—capabilities not contemplated or addressed by existing federal law.”
- Does the use of AI that “makes it possible to assemble this scattered, individually innocuous data into a comprehensive picture of any person’s life—automatically and at massive scale” raise Fourth Amendment concerns, in light of the Supreme Court’s precedent in Carpenter and Jones?
- Has the administration, including any of its departments or agencies, issued any legal analysis assessing whether the uses of AI technology for domestic surveillance are constitutional, such as those referenced in Kaplan’s declaration?
The Fourth Amendment’s “state action” doctrine applies when the government uses private companies as agents to obtain information. Anthropic’s complaint references the ability of Claude to engage in “collection and analysis of information,” suggesting DOD’s intended use may cross that threshold. Similarly, Kaplan’s declaration refers to “Claude collecting more information about U.S. persons than the user intended.”
- Does DOD plan to use Claude or any other AI company’s services to collect information, and not just to analyze it?
- Has the administration, including any of its departments or agencies, issued legal analyses assessing whether the collection of such information in this manner violates the Fourth Amendment or any other legal standard? What is the conclusion and the reasoning? Are there any limitations on AI use and, if so, what are they?
On Other Regulatory and Legal Standards
Even if constitutional, government purchasing of commercial information and analyzing it through powerful AI tools would require robust guardrails. (See, for example, Senator Wyden’s letter to former DNI Avril Haines on Jan. 25, 2024, referencing Federal Trade Commission (FTC) standards that may be relevant when purchasing commercially available information such as through data brokers.)
- What does DOD do to ensure that purchased commercially available information has been obtained with consumers’ consent and meets the standards that have been outlined by the FTC?
- What does DOD do with data that is found not to have been obtained with consumer consent or not to meet the relevant standards?
- Please provide the contracts and subcontracts with any companies involved in providing the DOD with commercial data.
In a letter to Congress on Jan. 15, 2021, the Defense Intelligence Agency (DIA) described its practices involving the use of commercial data. The letter states that DIA personnel have been granted permission to query the U.S. device location data “five times in the past two-and-a-half-years for authorized purposes.”
- Have any of the practices that DIA described changed since Jan. 15, 2021? Since Jan. 20, 2025?
- How many times were DIA or other DOD personnel granted permission to query the U.S. device location data from Jan. 20, 2021 to Jan. 19, 2025?
- How many times were DIA or other DOD personnel granted permission to query the U.S. device location data after Jan. 20, 2025?
- Do you commit to keeping Congress immediately informed if the rate ever increases significantly? (We note that Congress could and should adopt, as part of the reforms, a law mandating such disclosures.)
On OpenAI’s Contract and Government Surveillance
As of March 2, 2026, OpenAI has entered into a contract with the DOD for “deploying advanced AI systems in classified environments.” The contract provides: “For intelligence activities, any handling of private information will comply with the Fourth Amendment, the National Security Act of 1947 and the Foreign Intelligence and Surveillance Act of 1978, Executive Order 12333, and applicable DoD directives requiring a defined foreign intelligence purpose.”
- What does the administration believe are the conditions necessary to comply with the Fourth Amendment requirements of a “defined foreign intelligence purpose”?
- Does collecting information on or communications of “antifa” count as a foreign intelligence purpose?
The OpenAI contract also states: “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.” (emphasis added)
- What is the meaning of “unconstrained” as used in this contract in reference to the monitoring of U.S. persons’ private information?
- How is the statement in the OpenAI contract – that the system “shall also not be used for domestic law-enforcement activities except as permitted by the Posse Comitatus Act and other applicable law”– consistent with other provisions of the contract precluding use for domestic purposes? The latter provisions include:
“Consistent with applicable laws, including the Fourth Amendment to the United States Constitution, National Security Act of 1947, FISA Act of 1978, the AI system shall not be intentionally used for domestic surveillance of U.S. persons and nationals.”
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.”
- What is the meaning of “intentionally” as used in the provision on domestic surveillance? What (intentional or unintentional) uses of OpenAI would be permissible for domestic surveillance of U.S. persons and nationals under the contract?
On Changes to Legal and Policy Limits of Surveillance Authorities
- Since Jan. 20, 2025, has any department of agency altered guidance, policy, or practice on the collection or use of surveillance data on U.S. persons? If so, what are those changes?
- Since Jan. 20, 2025, has the administration adopted any new interpretation of the executive branch’s legal authority to target, collect, query or analyze the communications of U.S. persons?
- Since Jan. 20, 2025, has the Office of Legal Counsel provided any verbal or written opinion on the authority of the executive branch to target, collect, query or analyze the communications of U.S. persons?
- Has the FISA Court been asked to authorize any new surveillance programs, including the use of AI, since Jan. 20, 2025? If so, what are those programs? Has the FISA Court invalidated any proposed or enacted surveillance programs since Jan. 20, 2025?
- Has the Administration undertaken any mass surveillance without alerting the FISC?
- Why does the Administration need Section 702 if it is using AI and other surveillance tools? What specifically can it obtain using 702 that it cannot obtain through other tools at its disposal? Please provide specific examples.
On the President’s Article II Authority
- Does the administration believe it can conduct any form of foreign intelligence surveillance equivalent to Section 702 but under Article II authority without statutory authorization? If so, in what circumstances and subject to what constraints?
- Relatedly, is the administration’s view that the President has inherent Article II authority to engage in foreign intelligence surveillance equivalent to Section 702 if the Section 702 statutory authorities expire?
- Under Section 109 of FISA, it is unlawful for anyone to “engage[] in electronic surveillance under color of law except as authorized by statute.” 50 U.S.C. §1809(a)(1). Does the administration accept this provision is constitutional and binding on the executive branch?
- Has the administration conducted any warrantless surveillance since Jan. 20, 2025 under claimed inherent Article II authority? If so, what was the nature and scope of that surveillance, what targeting criteria were used, did it target or collect U.S. persons’ communications?
On Surveillance Operations Generally
- Has the administration engaged in collection or analysis of data on Americans involved in protests or other political activity opposing the administration’s immigration enforcement actions? Against political groups or parties? Against Members of Congress? Against candidates for office, state or federal?
- Since Jan. 20, 2025, has the administration targeted or collected the content of U.S. persons’ communications without a warrant? If so, under what circumstances? Under what circumstances, if any, does it believe it would be lawful to do so?
- Since Jan. 20, 2025, has the administration queried the content of U.S. persons’ communications without a warrant? If so, under what circumstances?
- In September 2025, the Trump administration issued National Security Presidential Memorandum-7, entitled “Countering Domestic Terrorism and Organized Political Violence” (NSPM-7), which was followed by the Attorney General’s implementing memorandum. Has NSPM been used by the Administration in support of any surveillance tools? (For more background, see Tom Joscelyn and Ryan Goodman, The Smearing of Alex Pretti and NSPM-7 (Jan. 26, 2026).)
