The Volokh Conspiracy

Mostly law professors | Sometimes contrarian | Often libertarian | Always independent

The Ninth Circuit's Stunner in Rosenow, and Thoughts on the Way Forward

A big problem in a small section, but one that can be corrected.


Two weeks ago, in United States v. Rosenow, the Ninth Circuit handed down a decision that has a brief but truly stunning passage with major implications for the future of Fourth Amendment law. If the passage stays in the opinion, it's going to have a quite dramatic effect, perhaps inadvertently, on how the Fourth Amendment applies to computers and the Internet.  And in my view, those effects are bad. Really, really bad. In this post, I want to say why the passage in Rosenow is so important, and why it's so disturbing.  And I'll suggest a possible path forward to deal with it.

I.  An Overview of the Rosenow Case

Rosenow arose out of an investigation into child exploitation in the Philippines.  Two large Internet providers, Yahoo and Facebook, ended up being involved.  That led eventually to Rosenow's arrest on his return from the Philippines to the United States and a search through his computers pursuant to a warrant. The search revealed child sexual abuse material (CSAM), also known as child pornography, including some involving sexual acts between Rosenow and prepubescent Filipina girls. On the basis of the images found on his computer, Rosenow was convicted of sexual exploitation and CSAM possession.

On appeal, Rosenow made several different arguments in his merits brief.  A majority of the merits arguments were on Fourth Amendment law.  The primary Fourth Amendment claim was that Yahoo and Facebook had become state actors, and that their accessing account information was a government search that triggered the Fourth Amendment.  But Rosenow also challenged other parts of the investigation, including the validity of the warrant obtained and whether accessing basic subscriber information was a search.

The Ninth Circuit affirmed in a published opinion by Judge Forrest, joined by Judge Callahan, with a partial dissent by Judge Graber.  The Court ruled that Facebook and Yahoo were acting as private actors, not state actors; that the warrants were valid, and that accessing basic subscriber information is not a search under the Fourth Amendment.  Judge Graber dissented in part as to one aspect of the state action ruling.

So far, this seems like an interesting decision, although on the whole relatively minor in its significance.  It's mostly fact-specific issues (such as where the state action line is) or else pretty clearly correct (such as the ruling that accessing basic subscriber is not a search).  It also has a lot I very much agree with, such as the specific ruling that the Stored Communications Act doesn't make Internet providers state actors.

But that's not all.  There's more. And that "more" turns out to be really big—much bigger in its importance, I think, than everything else in the Rosenow opinion.

II.  The Stunning Passage in Rosenow

What is that really big thing?  In his merits brief, Rosenow claimed that a set of other practices, including the preservation of his Internet account contents, violated the Fourth Amendment.  The argument is less than a page of the brief, see page 64, and the only material specifically about preservation is a single sentence saying that the preservations were seizures.

To my great surprise, Rosenow's merits opinion took on those claims and rejected them on the merits. And because Rosenow is a published opinion, its holding is now binding in the Ninth Circuit. In the 45-page published opinion, here's the totality of the discussion of preservation and the Fourth Amendment:

Were the preservation requests unconstitutional seizures?

Acting pursuant to 18 U.S.C. § 2703(f), which requires an ESP "to preserve records and other evidence in its possession pending the issuance of a court order or other process," the government directed Yahoo on three separate occasions to preserve records related to Rosenow's private communications. Rosenow contends that these requests were an unconstitutional seizure of his property.

A "seizure" of property requires "some meaningful interference [by the government,] with an individual's possessory interests in [his] property." Jacobsen, 466 U.S. at 113. Here, the preservation requests themselves, which applied only retrospectively, did not meaningfully interfere with Rosenow's possessory interests in his digital data because they did not prevent Rosenow from accessing his account. Nor did they provide the government with access to any of Rosenow's digital information without further legal process. It also is worth noting that Rosenow consented to the ESPs honoring preservation requests from law enforcement under the ESPs' terms of use. Thus, we agree with the district court that these requests did not amount to an unreasonable seizure in violation of the Fourth Amendment.


III.  Why the Passage Is So Important, and Why It's Really Bad.

Why is this passage so important?  There are two main reasons, I think.  The first is narrower, and the second broader.  Let me start with the narrower reason (although it's still pretty broad!).

First, as regular readers know, I think the Fourth Amendment makes most Internet content preservation unconstitutional.  You can see my full argument in this recent article of mine:  The Fourth Amendment Limits of Internet Content Preservation.  And you can see the argument fully written out as defense counsel might litigate it in this draft motion to suppress I recently wrote.  I think this is a hugely important issue, although it remained almost entirely unlitigated, much to my frustration.   No federal court of appeals has decided this issue until Rosenow.  The Ninth Circuit heard oral argument on this issue and had extensive briefing on it (with amicus oral argument participation, and briefing, from the ACLU) in United States v. Basey in 2019, but the panel did not reach the question in its decision.

For the Ninth Circuit to weigh in in this fashion when it wasn't fully briefed, and to reject these claims on the merits, is a big deal. It means that the government is free to pursue unlimited preservation, even though the major (and to my mind persuasive) arguments for why it shouldn't be able to do that have not been presented to the court that is saying this is permissible.

That's huge.  Under Rosenow, any government official can at any time order any Internet provider to run off a copy of anyone or even everyone's entire account contents.  The only authority on the question is now Rosenow, which rejects the claim. I think this is all wrong on the merits.  But more importantly, it's wrong on a vast scale: As explained in my article, there are hundreds of thousands of Internet preservations occurring every year.  This passage in Rosenow gives any government official a blank check to order any preservation of anyone or everyone's account without limit.  To decide this important a question in such a brief and unexplained passage is remarkable.

So that's the narrower (but again, pretty broad) problem.  Here's the even broader one.  The particular way that Rosenow rules for the government is incredibly far-reaching and deeply troubling.  Here's the key passage again:

A "seizure" of property requires "some meaningful interference [by the government,] with an individual's possessory interests in [his] property." Jacobsen, 466 U.S. at 113. Here, the preservation requests themselves, which applied only retrospectively, did not meaningfully interfere with Rosenow's possessory interests in his digital data because they did not prevent Rosenow from accessing his account.

This is a remarkably far-reaching ruling, and all in just two sentences.  Whether and when copying data is a Fourth Amendment seizure is one of the foundational questions of the future of Fourth Amendment law that has been in the background of the caselaw for years.  And it's also been a topic in scholarship for a long time, including stuff I have spent a lot of time on. As longtime readers may remember, in 2005 I initially concluded that copying wasn't a seizure.  By 2010, however, I had changed my mind, and I wrote an article explaining why copying ordinarily is a seizure, using preservation requests as an example: Fourth Amendment Seizures of Computer Data, 119 Yale Law Journal 700 (2010).

District courts have divided on the question.  And the one federal court of appeals ruling on it, which had held that copying is a seizure, was vacated when the court went en banc (presumably for reasons other than the seizure ruling, as there was a lot more going on in the case).  See United States v. Ganias, 755 F.3d 125, 137 (2d Cir. 2014) (holding that the Government's retention of electronic copies of the defendant's personal computer "deprived him of exclusive control over those files," which was "a meaningful interference with [the defendant's] possessory rights in those files and constituted a seizure within the meaning of the Fourth Amendment."), vacated by United States v. Ganias, 824 F.3d 199 (2d Cir. 2016) (en banc).  So until Rosenow, there was no circuit court precedent on this foundational question.

By ruling that copying is not a seizure, the Ninth Circuit becomes the first circuit court to have ruled on the question in a decision that is still on the books.  And the question is of dramatic importance, as the introduction to my 2010 article flags:

Imagine the police take away a suspect's computer, make a digital copy of its contents, and then give the computer back to the suspect. The police do not open the copy, but they keep it in their custody in case they need to access it later. Does the combined act of copying the files and retaining the copy trigger the Fourth Amendment?

Next imagine that FBI agents believe a particular person is using the Internet to commit a crime. Agents install a surveillance tool at the target's Internet service provider (ISP), and the tool generates copies of all of the target's incoming and outgoing email. The email is copied to a file, but no human being actually looks at the file. Instead, the agents keep the file in case they develop probable cause to look through it for evidence. Again, does the Fourth Amendment allow it?

The answer to both scenarios depends on whether copying computer files without looking at them constitutes a Fourth Amendment "seizure.  If copying a computer file amounts to a seizure, then the government cannot make and retain a copy absent special circumstances. On the other hand, if copying is not a seizure, then the government can make and retain the copy without restriction. The Fourth Amendment will limit looking through the copy because that is a Fourth Amendment "search. But what if the government wants to make a copy and hold it? Does that constitute a "seizure"?

Putting the preservation issue aside, a ruling that copying is not a seizure has extraordinary implications.  Whether you agree with that holding or disagree with it, I don't think it's a conclusion that is best reached in a precedential opinion in just two sentences.

And that's not all.  The Rosenow passage includes this sentence:

It also is worth noting that Rosenow consented to the ESPs honoring preservation requests from law enforcement under the ESPs' terms of use.

What were those terms of use?  Earlier in the decision, the court notes a fairly standard term in its privacy policy, here Yahoo's, that Yahoo may disclose information "to investigate, prevent, or take action regarding illegal activities . . . , violations of Yahoo's terms of use, or as otherwise required by law."  I assume that's the term they had in mind.

I'm not entirely sure how to read that "note." But read fully, it seems to be suggesting that agreeing to terms of use governing an Internet account can eliminate Fourth Amendment rights.  In particular, the provider telling you in the fine print when you register the account that they might end up giving your data to the government in some situations is treated as forever blanket consent to that as a matter of Fourth Amendment law.

As a matter of Fourth Amendment law, I think that would be wrong, for reasons I argued here.  But it's also about as far-reaching as you can get.  If agreeing to standard terms of service is consent to a possible seizure, then presumably it is also consent to a possible search.  That would mean that Fourth Amendment rights online can be reduced to zero through a standard online term no one reads, and that users can't avoid, that were created just to make sure no one can sue the providers for complying with valid legal process.  I don't want to be apocalyptic, but this seems really really bad.

IV.  A Possible Way Forward

What are the prospects for further review?  En banc review is possible. Counsel for Rosenow has received an extension for a petition for rehearing, until June 8.  When the petition for rehearing is filed, it's possible that the court could want an en banc panel to take a look at the preservation issue, which could lead to a new opinion. And I would certainly support a full look at the preservation issue by the en banc court.

But there's also a narrower way, should the court not want to go en banc on this or another part of the case.  It seems to me that another way forward would be for the panel to just amend the opinion, only changing that passage, to say the issue will not be addressed because it was not fully briefed and was therefore waived.

Here's some context.  Under Ninth Circuit practice, an issue is deemed waived unless the appellant "specifically and distinctly argue[d] the issue in his or her opening brief." United States v. Kama, 394 F.3d 1236, 1238 (9th Cir. 2005). The appellant has to specifically and distinctly raise the issue, and it then has to be "supported by argument" in the opening brief, for the Ninth Circuit to rule on it. Leer v. Murphy, 844 F.2d 628, 634 (9th Cir.1988).

As I read the opening brief, whether preservation is an unreasonable seizure was neither specifically and distinctly argued nor supported by argument in the opening brief.  Here's the full section from the opening brief:

The government's subpoenas and preservation requests were also illegal searches and seizures under Carpenter.

The recent Supreme Court case Carpenter v. United States, 138 S. Ct. 2206 (2018) also instructs that Rosenow had a legitimate right to privacy in his digital data, and that it violated the Fourth Amendment to interfere with that right without a warrant and probable cause. In Carpenter, the government obtained orders directing wireless carriers to provide cell-tower data regarding several criminal suspects. Id. at 2212. The Supreme Court reversed the resulting conviction, holding that warrantlessly obtaining this information violated the Fourth Amendment. In so doing, it rejected the notion that the third-party doctrine insulated this information from Fourth Amendment scrutiny, noting that thirdparty- doctrine cases did not deal with "confidential communications" and other private information. Id. at 2219. [FN158] The Court held that a warrant should have been required: "this Court has never held that the Government may subpoena third parties for records in which the suspect has a reasonable expectation of privacy" it observed. Id. at 2221 (emphasis provided). "If the choice to proceed by subpoena provided a categorical limitation on Fourth Amendment protection, no type of record would ever be protected by the warrant requirement." Id. at 2222.

Carpenter demonstrates that searches and seizures occurred here. The government seized Yahoo records through ongoing preservation requests, with no notice to Rosenow. And it both seized property and affirmatively prompted additional searches by issuing administrative subpoenas to Facebook. Under Carpenter, this should have required a warrant showing probable cause. Because the government had neither, this evidence should have been suppressed.

[FN158: Even the dissent seemed to concede that private communications—as opposed to mere location data—would not be governed by the third-party doctrine. See id. at 2230 (Kennedy, J., dissenting) ("Miller and Smith [the leading third-party cases] may not apply when the Government obtains the modern-day equivalents of an individual's own 'papers' or 'effects,' even when those papers or effects are held by a third party.") (citing Ex parte Jackson, 96 U. S. 727, 733 (1878) (letters held by mail carrier); United States v. Warshak, 631 F. 3d 266, 283-288 (6th Cir. 2010) (e-mails held by Internet service provider)).]

The claim here seems to be that, somewhere in the subpoenas and preservation requests, there were unreasonable searches and seizures.  Most of the passage seems to be arguing that the subpoenas were searches under Carpenter, a claim that in my view the Ninth Circuit properly rejected.

But here's the key.  As I read that passage, it doesn't seem to be separately and distinctly claiming that preservation is an unreasonable seizure, or offering a legal argument for why that might be.  The test for what is a seizure is not even mentioned. How the test might apply is not mentioned.  The test for the reasonableness of a seizure is not mentioned (much less applied) either, except to say that a warrant should have been required.  Compare that passage to the model brief I recently posted. It seems to me that the opening brief in Rosenow did not put the court on notice of the legal questions before it on the preservation issue.  You'd have to guess what the specific arguments were on the preservation issue, as the opening brief didn't flag them, much less make them.

I don't mean to criticize the hard-working counsel in the case. I believe Rosenow was the first case ever to get an opinion on the Fourth Amendment issues raised by preservation in the trial court decision.  To use an exam-grading analogy ('tis the season): The issue was spotted, unlike all the other cases where counsel have just missed it.  The opening brief raised a lot of issues, and went right up to the page limit, so it may not have been developed because of other issues in the case.  And I agree that the preservation issue was addressed in more detail in the Ninth Circuit reply brief.

But it seems to me that, if the court doesn't want to take this enormously important set of questions on en banc, it might be wise to amend the panel opinion to recognize that there's a lot more going on in the case than the panel realized given that the issue was not fully developed in the opening brief.  That would make no difference to the outcome. And it would just replace a single paragraph in the opinion, as the analysis paragraph would just become a waiver paragraph.  But it would mean that these critical and novel legal issues could wait for full briefing and decision in another case.

As always, stay tuned.

[UPDATE: I fiddled a bit with the post shortly after posting it.]