This week, Just Security ran two posts, one by Matt Olsen and one by Asha Rangappa, defending the government’s warrantless access to Americans’ communications obtained “incidentally” under Section 702. Both authors take a “nothing to see here” approach, analogizing the practice of backdoor searches to other standard investigative techniques. But their comparisons are rife with logical flaws and basic factual inaccuracies. In fact, the Section 702 regime flies in the face of criminal practice and precedent. We write here—both of us as longtime civil liberties advocates, and one of us as a 16-year veteran of the FBI—to set the record straight.
Section 702 and Backdoor Searches
Let’s start with a more complete description of Section 702 surveillance. In discussing the “lawful targets” of Section 702 surveillance, Olsen and Rangappa mention only suspected terrorists and malign foreign powers as examples. But Section 702 allows surveillance of any foreigner overseas, as long as a significant purpose is to obtain “foreign intelligence”—a term defined to include information that merely “relates to” the conduct of foreign affairs. This loose definition matters because it means that the Americans whose communications are “incidentally” collected are not necessarily communicating with terrorists or Russian agents. They may well be communicating with a foreign friend or relative who is not suspected of any wrongdoing whatsoever.
Olsen and Rangappa also say nothing about the mindboggling scale of collection. Under Section 702, the NSA obtains hundreds of millions of Internet communications each year. While the government refuses to provide any estimate of how many of these communications involve Americans, it is safe to assume that the total number of Americans’ communications “incidentally” collected since the inception of Section 702 is well into the millions.
Of course, this warrantless collection of Americans’ communications would be unconstitutional if the government’s purpose was to spy on the Americans, rather than the foreigners with whom they are communicating. The government must accordingly certify to the FISA Court that its intent is not to access any particular, known American’s communications. And yet, immediately upon obtaining the data, government agencies (including the FBI, the NSA, the CIA, and the NCTC) search through it for the communications of particular, known Americans. This is the practice known as “backdoor searches.”
Incidental Collection and Access to “Lawfully Collected” Data
Olsen defends this practice by analogy to electronic surveillance in ordinary criminal cases. He claims that incidental collection under Section 702 is no different from the incidental collection of communications of non-targets in the context of criminal wiretaps. For example, when a suspected mobster uses a wiretapped phone to order a pizza, both sides of the communications may be monitored and recorded, even though the pizza delivery person was never the subject of a court-authorized wiretap . . . This is beyond dispute as a matter of constitutional law.
This is flatly wrong. Under longstanding Supreme Court precedent, a criminal wiretap must specify not only the phone line to be tapped, but also the conversations to be acquired. Although the government may collect conversations about the suspected crime that involve people not named in the warrant, it must take affirmative steps to avoid acquiring any conversations that do not relate to criminal activity. In order to fulfill this requirement, the government employs real-time minimization practices, which means an agent sits and listens to the intercepted communications as they happen, and turns off the recording device whenever non-pertinent conversations take place. For stored communications such as e-mails, a supervising attorney reviews them and passes along only the pertinent material to investigators.
Thus, in stark contrast to Section 702 surveillance, the government in criminal cases is not “free to seize at will every communication” of the target. It would be unconstitutional for the government to deliberately record a telephone conversation between the suspect and the pizza delivery person unless it was apparent that the pizza guy was in on the crime.
In any case, the question is not whether the practice in criminal cases justifies the incidental collection of Americans’ communications under Section 702 (it doesn’t), but whether a warrant should be required to access the incidentally collected communications. Here, Olsen theorizes that a query of collected information “is not the initiation of a new surveillance or search protected by the Fourth Amendment and subject to the warrant requirement.” But that’s exactly what it is—as explained here by Orin Kerr, a leading Fourth Amendment scholar whose analyses are frequently cited in Supreme Court opinions.
Olsen also argues that there is “simply no precedent” for requiring a warrant to search information that has been lawfully collected. This, too, is wrong. In one recent case, for example, a law enforcement agency was required to obtain a warrant in order to search data that had been lawfully collected by a different law enforcement agency for a different purpose. The court imposed this requirement even though the data was originally acquired pursuant to a warrant—i.e., with far more protection for “incidentally” collected data than is present in Section 702 collection. In another case (ultimately reversed on other grounds), the court went even further, holding that the government was not entitled to conduct a second search of lawfully collected data even when it did obtain a separate warrant, because the government should not have retained the data after the original purpose for collecting it no longer existed.
Indeed, restrictions on searches of lawfully accessed property—whether physical or digital—are the constitutional norm, not the exception. For instance, if a police department obtains a warrant to search a house for a murder weapon, agents can lawfully access the house and search every room. But after finding (or failing to find) the murder weapon, they can’t go rifling through file cabinets for evidence of tax evasion, even though the officers are lawfully present in the house. They would have to get a separate warrant to conduct that search.
Olsen correctly observes that the FISA Court has upheld the constitutionality of backdoor searches. So have two or three district courts, depending on how one reads an ambiguous footnote in one of the decisions. One of us has already detailed the flaws in the FISA Court’s opinion (from which no appeal is possible), so we won’t repeat them here.
As for the district courts, Olsen quotes their rationale: “It would be perverse to authorize the restricted review of lawfully collected information but then restrict the targeted review of the same information in response to tailored inquiries.” He fails to note the critical context for this statement, i.e., the “restricted review” to which the courts were referring. The Department of Justice had informed the courts that the government was already required to review Americans’ incidentally obtained communications in order to apply minimization procedures. As stated in one of the opinions:
The government notes the minimization procedures compel it to review information lawfully collected under § 702, which includes information about U.S. persons, to determine if the information should be retained or disseminated. According to the government:
It would be perverse to authorize the unrestricted review of lawfully collected information but then to restrict the targeted review of the same information in response to tailored queries. Querying lawfully collected information using U.S.-person identifiers does not involve a significant additional intrusion on a person’s privacy, beyond the level of intrusion already occasioned by the government as it reviews and uses information it lawfully collects under Section 702 pursuant to its need to analyze whether the information should be retained or disseminated.
The court relied on this representation by the Justice Department in reaching its conclusion that backdoor searches would not add any new privacy intrusion. But the representation was false, as Olsen surely knows. None of the agencies’ minimization procedures require them to review Americans’ communications obtained under Section 702 to determine whether they should be kept or disseminated, and none of the agencies conducts such a review. Indeed, in resisting calls to provide an estimate of how many Americans’ communications are obtained under Section 702, intelligence officials have argued that the undertaking would violate Americans’ privacy because it would require the government to review Americans’ communications, which would probably never be reviewed in the ordinary course of business. (The claim that providing an estimate would require reviewing Americans’ communications is inaccurate, but the observation that most of these communications are never reviewed is correct.)
In short, existing practice and case law in the criminal context support a warrant requirement for searching Americans’ communications obtained under Section 702. Those who argue otherwise, including the three or four trial-level courts that have upheld backdoor searches, are relying on a combination of flawed logic and erroneous facts.
Are Warrants a Threat to National Security?
Olsen and Rangappa both construct hypothetical scenarios in an effort to show that requiring a warrant for backdoor searches would prevent the government from discovering terrorist plots. The use of hypotheticals here is extremely revealing. While the government has provided multiple public examples of Section 702 surveillance being used to thwart terrorists, there is no indication that any of these success stories involved backdoor searches. Indeed, the government has conspicuously failed to assert, even in the most general terms, that backdoor searches have provided significant counterterrorism benefit. Instead, after several years of the FBI conducting backdoor searches as a matter of routine, the government is still forced to rely on the argument that backdoor searches might prove useful in the future.
In any case, the hypotheticals they provide don’t support the claim that a warrant requirement would stymie terrorism investigations. Take Olsen’s scenario, in which a tipster tells the FBI that his neighbor is talking about wanting to work with ISIS. Buried within the Section 702 data is an e-mail between the neighbor and an “ISIS handler,” in which the handler instructs the neighbor to drive a truck into a crowd. A warrant requirement for US person queries, Olsen suggests, would prevent the government from ever learning about this plot.
But the government presumably would have discovered this e-mail already, in the course of monitoring the target—i.e., the ISIS handler. Olsen’s hypothetical makes sense only if the government is not bothering to review the communications of ISIS handlers that it intercepts. If that is the case, the national security threat in this scenario is not the warrant requirement, but the government’s extreme negligence in conducting “surveillance” of ISIS handlers that consists of collecting their communications and stashing them away, unreviewed, until someone calls in a random tip about a neighbor.
This brings us to Olsen and Rangappa’s dutiful recitation of the claim that requiring a warrant would be akin to reconstructing “the wall.” This term describes perceived restrictions on the sharing of information between foreign intelligence and law enforcement officials in the years before 9/11. In fact, not a single Section 702 reform proposal currently before Congress—not even the most ambitious version, authored by Senators Ron Wyden and Rand Paul—would limit the sharing of terrorism-related information acquired under Section 702. If, in the course of reviewing an ISIS handler’s e-mails, the NSA came across an e-mail urging an American to drive a truck into a crowd, there would be no restriction whatsoever on the NSA’s ability to share the e-mail with the FBI.
Olsen and Rangappa’s real complaint is not that a warrant requirement would resurrect the pre-9/11 obstacles to sharing terrorism information (it wouldn’t), but that warrants are themselves obstacles to accessing information about Americans. That is surely true, but it’s a constitutional feature, not a bug. Contrary to the authors’ supposition, advocates of a warrant requirement do not “misunderstand” the nature of FBI investigations. We are fully aware that “criminals and spies don’t show up on the doorstep of law enforcement with all of their evidence and motives neatly tied up in a bow,” and that the FBI may lack anything close to probable cause in the early stages of a case. Lack of evidence that an American is engaged in wrongdoing, however, is hardly a compelling justification for a warrantless search of his or her communications. To the contrary: if the FBI does not have probable cause to suspect criminal activity, it has no business reading Americans’ e-mails and listening to their phone calls.
This does not leave the FBI powerless to investigate, by any stretch. The FBI has a wide range of potent investigative tools that have expanded significantly since 9/11, as one of us wrote about here. Rangappa herself observes that agents can interview witnesses and contacts, dig through trash, use undercover informants, and obtain third-party records without meeting probable cause requirements. Rangappa argues that these pre-warrant tactics are “even more aggressive and intrusive” than accessing and reading the contents of the suspects’ e-mails, but this claim is clearly wrong as a matter of constitutional law as well as common sense. From a privacy perspective, reading the contents of an Americans’ private communications is among the most intrusive activities possible, which is why the government must obtain a warrant to do it.
It must also be noted that “probable cause” is not the dauntingly high bar that Olsen and Rangappa portray. It is simply the evidence necessary, under the totality of the circumstances, to convince a reasonable law enforcement officer that there is a fair probability that a crime is being or has been committed, and that evidence of that crime can be found in the place to be searched. Many thousands of search warrants and wiretap orders are issued each year in criminal cases as serious as murders, rapes, and kidnappings, as well as run-of-the-mill burglaries and petty thefts. None of these cases begin with criminals “show[ing] up on the doorstep of law enforcement with all of their evidence and motives neatly tied up in a bow.” Yet somehow, the government manages to obtain the evidence it needs to protect our safety, while also respecting the constitutional limits necessary to protect our individual rights and privacy.
Finally, Olsen is off base when he asserts that “there is no principled or logical limit to the view of critics that the FBI should have to establish probable cause and get the permission of a judge to look at information it has already obtained lawfully.” The limiting principle is clear and easily administrable: the government should not be able to access information that is entitled to Fourth Amendment protection, such as the content of an American’s communications, without first having obtained a warrant. This principle would place no limits on searches of information, such as voluntarily given witness statements (mentioned by Olsen), that does not merit Fourth Amendment protection; nor would it limit searches of information obtained pursuant to a warrant, as long as the search falls within the warrant’s parameters.
What the Constitution cannot tolerate is the government’s routine mining of Section 702 data, obtained without a warrant based on a promise that only foreigners would be targeted, for the communications of specific Americans. There is no analogy or precedent for such a practice anywhere outside the four corners of Section 702. It is time to end it.