In March, the Brennan Center published a two-page document on its website setting forth myths and facts about “backdoor searches”— the government’s practice of searching through communications acquired without a warrant under Section 702 of FISA for the purpose of finding Americans’ phone calls, text messages, and emails. The Brennan Center and other critics of the practice have long urged Congress to require the government to obtain a warrant or FISA Title I order before accessing Americans’ communications collected under Section 702.
George Croner, a longtime opponent of this reform, responded to the Brennan Center’s Myths and Facts document in a lengthy Just Security piece that took issue with several of our points. For the most part, his piece retreads well-worn ground—e.g., claiming that warrantless backdoor searches are needed for national security when an 18-year track record says otherwise. We write here simply to highlight four places where his analysis is inaccurate or incomplete.
Are backdoor searches constitutional?
The Brennan Center’s brief document mentioned one 2024 district court case (United States v. Hasbajrami) holding that the government must obtain a warrant or cite an applicable exception to the warrant requirement to conduct backdoor searches. Croner responds that this case is an “outlier” and that “one district court’s (faulty) ruling cannot displace nearly two decades of review by the FISC and FISCR.”
To start, this statement leaves out important context. The Foreign Intelligence Surveillance Court (“FISC”) and its appellate body, the Foreign Intelligence Surveillance Court of Review (“FISCR”) (collectively, the “FISA Court”), are not regular courts. They normally hear only from one party: the government. Even when amici curiae participate, they are limited in their access to materials and cannot appeal adverse rulings by the FISC. The government attorneys who appear before the courts are a small cadre of repeat players who are routinely in contact with the court staff over substantive matters. The rulings of these specialized courts are rarely tested outside of this closed system.
It should be no surprise that this legal echo chamber tends to produce rulings in the government’s favor. Notably, the FISC also signed off — multiple times — on the NSA’s bulk collection of Americans’ phone records; once that program became public, the three regular federal courts that reviewed it all parted ways with the FISC, with two holding that the program was unlawful (one of these decisions was later reversed on standing grounds) and one expressing doubt about its lawfulness.
Unfortunately, regular courts have had extremely few opportunities to weigh in on Section 702, and even fewer to address backdoor searches. That is in large part because the government for years violated a statutory requirement to notify criminal defendants when using evidence derived from Section 702 collection.
Among the few regular federal courts that have confronted backdoor searches, the district court in Hasbajrami is not an “outlier.” In fact, that decision followed a unanimous ruling by a three-judge panel of the Second Circuit Court of Appeals — the only regular federal appellate court that has squarely ruled on the issue — rejecting the FISA Court’s rationale, i.e., that backdoor searches are not separate Fourth Amendment events from the initial collection and therefore do not trigger an independent Fourth Amendment analysis. In another case, a judge on a three-judge panel of the Tenth Circuit agreed with the Second Circuit that querying constitutes a separate Fourth Amendment event that must be independently evaluated. (The majority in that case acknowledged that querying “might raise difficult Fourth Amendment questions,” but it did not reach them, accepting the government’s assertion — over the dissent’s objection — that the evidence in the case was not obtained through backdoor searches.)
The Supreme Court has consistently admonished that, where the Fourth Amendment is implicated, warrantless searches are “per se unreasonable . . . subject only to a few specifically established and well-delineated exceptions.” Accordingly, if backdoor searches require a separate Fourth Amendment analysis, it follows — and the district court in Hasbajrami ultimately held — that the government must either obtain a warrant or cite an applicable exception to the warrant requirement to conduct them. Although the district court also held that a so-called “foreign intelligence exception” would authorize some backdoor searches (albeit not the ones in that case), this aspect of the court’s analysis was flawed for reasons one of us explained here. (To the extent Croner suggests that the FISA Court has relied on the foreign intelligence exception to uphold backdoor searches, that is incorrect; because the FISA Court does not treat backdoor searches as separate Fourth Amendment events, it has concluded that they need not qualify for any exception to the warrant requirement.)
In short, of the four regular federal appellate court judges to address backdoor searches, all four have rejected the FISA Court’s approach. True, an equal number of district court judges reached the opposite conclusion (with most relying heavily on a misrepresentation by the Department of Justice about how minimization works, as one of us explained here). But in the regular court system — unlike the FISA Court mechanism — there are 94 district courts and 14 appellate courts (including the Supreme Court), with appeals generally available to both parties. This system ensures that errors made by district courts can be corrected as cases wend their way upwards.
Did RISAA solve the problem?
Opponents of a warrant requirement suggest that the reforms included in the 2024 Reforming Intelligence and Securing America Act (“RISAA”) are sufficient because the government has reported a steep decline in the FBI’s backdoor searches and an increased rate of compliance with internal standards. Even if these statistics were complete and accurate, they would not obviate the need for a warrant. As Chief Justice Roberts wrote in a seminal privacy case, “The founders did not fight a revolution to gain the right to government agency protocols.”
In any event, the statistics for 2024 rely on incomplete data, as the Brennan Center pointed out in its Myths and Facts document. That is because the FBI did not track or audit queries that agents performed using a querying tool known as an “advanced filter function.” Indeed, the government acknowledged that the FBI likely did not follow any of RISAA’s procedural requirements for U.S. person queries when using this tool. This systemic violation of multiple provisions of RISAA on its own rebuts any claim that RISAA solved the FBI’s compliance problems. (We discuss this issue further in our recent Just Security post.)
The government recently reported that the FBI conducted 7,413 searches in 2025. Croner asserts that this number — which is also markedly lower than numbers in previous years — was “largely unaffected” by the FBI’s violations because the advanced filter function was discontinued in “early 2025.” But the government’s reporting periods cover December through November, and “early 2025” does not necessarily mean January 1. The actual number of queries in 2025 could be significantly higher than the reported number of 7,413.
The more fundamental question is whether the government’s reporting for 2025 can be trusted. The FBI already operated one querying tool under the radar, and the transgression was reported to the FISC only because the Justice Department’s National Security Division (“NSD”) discovered it. The FBI had to engage in serious interpretive gymnastics to justify its position that searches using the advanced filter function fell outside the statutory definition of “queries.” It would be willful blindness not to consider the possibility that the FBI has developed other, similarly creative workarounds that the NSD has yet to discover or report — such as using artificial intelligence to analyze U.S. person communications.
More broadly, to rely on the government’s statistics or compliance reports for 2025, one must assume that the current Department of Justice (including both NSD and FBI) is scrupulously tracking and auditing queries and accurately reporting the results to facilitate oversight by Congress and the courts. There is ample reason to question this assumption. The administration has gutted the main oversight mechanisms for Section 702, dismantling the Office of Internal Auditing (the office established in 2020 to improve compliance with Section 702), firing all three Democratic appointees on the five-member Privacy and Civil Liberties Oversight Board (thus undermining both its effectiveness and its independence), and apparently cowing DOJ’s Inspector General into inactivity. And dozens of courts across the country have admonished DOJ for providing inaccurate, incomplete, or misleading information.
Would a warrant requirement create an unworkable burden on the courts?
Croner calculates that a warrant requirement might require the FBI to seek more than 2,000 warrants each year. He bases this number on the fact that the FBI reported 7,413 backdoor searches in 2025, 28% of which returned either metadata or communications content. (The leading reform proposals would require the government to obtain a warrant after performing the query but before reviewing any returned content.) If the actual number of queries in 2025 was higher than 7,413, as it almost certainly was due to the use of the advanced filter function, the number of queries leading to warrant applications could in theory be larger.
In practice, however, the number would almost certainly be much lower. According to the former head of NSD, “[i]n most cases, the government is conducting queries in the earliest moments of an investigation, reviewing the data in its possession at a time when there may be little available information about a potential threat.” The existence of a warrant requirement would put an end to such fishing expeditions, as the government would not seek a warrant in cases where it had little to no evidence of wrongdoing.
That is a feature of the reform, not a bug. The government should not be attempting to access Americans’ private communications with nothing close to probable cause. The FBI has plenty of less intrusive investigative tools at its disposal to build a case and obtain the evidence necessary to support a warrant application.
If Section 702 were to expire temporarily while Congress is considering reforms, would Section 702 surveillance be hindered?
Section 702 operates under year-long “certifications” that authorize surveillance to acquire broad categories of foreign intelligence. Croner acknowledges that the government would be legally authorized to continue Section 702 surveillance under existing certifications (which will remain in force until spring of 2027) even if the underlying law were to expire. However, he raises the specter that companies might nonetheless “stop complying with Section 702 directives.”
Companies do not choose whether to assist the government with Section 702 surveillance. A company that defies a valid directive faces fines of $250,000 per day. And the validity of a directive is determined by the FISC, not the company. In 2008, Yahoo refused to comply with a directive during a brief lapse between the expiration of Section 702’s predecessor and the enactment of Section 702. The FISC confirmed that the directive remained valid and ordered Yahoo to comply. Congress subsequently strengthened the language of the grandfathering provision to ensure that there would be no repeat of this incident.
While Croner is correct that the government would not be able to obtain new certifications during this period, he is off base in claiming that “it is very easy to imagine a severe need for a new certification.” Croner doesn’t cite a single historical example of a severe need arising outside of the annual certification process, and we are not aware of one. (The government sought new certifications after RISAA’s passage because RISAA required them.) Existing certifications are used to combat terrorism, foil malicious cyberactivity, disrupt international drug trafficking, counter the proliferation of weapons of mass destruction, and conduct surveillance of foreign governments and related entities. Indeed, as Croner observes, existing certifications allow surveillance for every single NSA-reported topic in the president’s intelligence priorities. It is a challenge, to say the least, to imagine the “severe new need” Croner posits.
***
Pending judicial resolution of the Fourth Amendment issues at stake, Congress faces a choice about whether to require warrants for backdoor searches. Members will have different views on this matter. But those views should be rooted in a shared, accurate understanding of the facts — not the pervasive myths that have too often clouded the debate.
