Fourth Circuit Takes a Wrecking Ball to Zeran and Section 230–Henderson v. Public Data

Ugh, this Fourth Circuit opinion is terrible. I instantly knew it was written by a Trump appointee from the Federalist Society because it had the characteristic self-assured yet sloppy writing style. And what’s the deal with Justice Thomas fandom among those judges? This opinion gives 4 gratuitous cites to Thomas dissents.

This opinion does extensive damage to Section 230, including seemingly overwriting core parts of the seminal Section 230 case, Zeran v. AOL. The associated chaos will likely continue until the Supreme Court’s Gonzalez decision, which will likely do so much greater damage to Section 230 that this opinion’s troubles will seem minor in comparison.

Overview

The plaintiff sued the defendant for violating four provisions of the Fair Credit Reporting Act (FCRA). At its core, the FCRA is in tension with Section 230 because it seemingly regulates the dissemination of third-party content (i.e., the credit data provided by vendors). However, many FCRA provisions are ministerial in nature about how to operate a credit reporting agency, and those provisions may not specifically base liability for credit report dissemination even if the overall statutory objective relates to that output. This makes the FCRA/230 interface ambiguous and confusing.

The district court dismissed the lawsuit on Section 230 grounds in a garbled and controversial opinion. As I predicted then, the district court’s “distorted Section 230 test makes this ruling vulnerable on appeal or in further proceedings.” And here we are.

The court says that 230 could apply to FCRA claims, but it didn’t in this case because two of the claims did not treat the defendant as a publisher/speaker and two other claims were based on deceptive omissions and summaries created by the defendants that were not in the source material. While that top-line assessment may not sound too bad, the opinion’s sloppiness and overreach cause much more damage to Zeran and 230 than was required to achieve its results.

Publisher/Speaker Claims

Citing Zeran, the court says the term “publisher” derives from traditional defamation law. (The opinion engages in selective textualism by completely ignoring the “speaker” reference in 230(c)(1)). Per defamation law, the court says publishers disseminate information to third parties (so any information disseminated only to the plaintiff couldn’t be defamatory). Further, defamation only imposed liability when the disseminated information was wrongful. This leads to a superficially innocuous conclusion:

for § 230(c)(1) protection to apply, we require that liability attach to the defendant on account of some improper content within their publication

Say what? That’s expressly NOT what Congress said. 230(c)(1) says: “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider” (emphasis added). The statute is clear: it provides immunity for “any information,” not “any improper information.” Thus, the court manufactures a new non-textualist statutory requirement.

If the court means what it says, this opens up the door for  plaintiffs to claim they are suing for something other than “improper content within their publication.” We’ve seen so many cases unsuccessfully try this workaround to 230 in the past, and this court now gives those plaintiffs a new attack vector to reframe their complaint and sidestep 230 entirely.

Indeed, consider how the court’s standard would have applied to Zeran. One of Zeran’s arguments was that AOL was negligent for failing to keep the information he needed to identify the wrongdoer. Sounds like Zeran’s argument would have gotten around 230 if this opinion had applied.

The court reject the defendant’s argument that 230 applies any time publication is a “but-for” element of the claim. (Citing Erie Insurance, the court says: “This but-for test bears little relation to publisher liability at common law”). The court summarizes its analysis:

a claim only treats the defendant “as the publisher or speaker of any information” under § 230(c)(1) if it (1) bases the defendant’s liability on the disseminating of information to third parties and (2) imposes liability based on the information’s improper content

So, do the plaintiff’s claims treat the defendant as a publisher/speaker? The court says that “we must look beyond the claim’s formal elements” and instead use “a functional approach”:

  • FCRA § 1681g requires credit reporting agencies (CRAs) to provide consumers with their credit reports. Because these disclosures go to the plaintiff, not third-parties, the court says it’s not publishing and therefore is categorically outside 230(c)(1).
  • § 1681b(b)(1) requires CRAs to take some steps before furnishing credit reports for employment purposes. First, the CRA must obtain certification from employers. The court says this only imposes liability for “failure to obtain the required information (certification) from the employer,” not third-party content. Second, the CRA must inform employers of the consumer’s rights. This is excluded from 230 because the communicated information “is proper and lawful content,” which doesn’t satisfy the court’s new non-textualist “impropriety” test.
  • Section 1681e(b) requires CRAs to confirm the accuracy of information it has about consumers. § 1681k(a) requires CRAs to provide complete and current information when providing credit reports for employment purposes. Both impose liability based on the content’s impropriety, and plaintiffs will only have standing when the “improper content” is disseminated to third parties (TransUnion v. Ramirez). The court punts on this issue because it says that even if these are publisher claims, they aren’t based on third-party content.

Provided by Another Information Content Provider

The court takes a Roommates.com-style approach that a defendant partially develops the content when it adds illegality to third-party content (the so-called “material contribution” test). Stated this way, the opinion shouldn’t do much extra damage beyond Roommates.com. And yet…

The court summarizes its legal standard as:

an interactive computer service is not responsible for developing the unlawful information unless they have gone beyond the exercise of traditional editorial functions and materially contributed to what made the content unlawful.

This adopts an obvious false dichotomy. Materially contributing to third-party content is a “traditional editorial function,” so this distinction is incoherent. This legal standard invites plaintiffs to define “traditional editorial functions” to exclude whatever defense behavior they are targeting. The chaos is palpable. (The scope of “traditional editorial functions” is a question presented in Gonzalez, so the Supreme Court will almost certainly make a worse hash of this term by June).

Applying this standard, the court says that the defense added the illegality:

[the plaintiffs] sufficiently allege that the inaccuracies in McBride’s report resulted from Public Data’s stripping out the nolle prosequi disposition for McBride’s charges and adding in its own misleading summaries.

Thus, on Plaintiffs’ allegations, Public Data’s summaries and omissions materially contribute to the report’s impropriety. They are not merely an exercise of traditional editorial functions….

as alleged, the content provided to Public Data about McBride was not inaccurate. Instead, through Public Data’s actions, the records were changed so as to introduce the inaccuracies. Public Data thus made substantive changes to the records’ content that materially contributed to the records’ unlawfulness.

Can you see the looming problem? When a 230 defendant republishes third-party content verbatim and without redaction, this standard is fine. But 230 defendants routinely extract pieces of a third-party submission–sometimes as promotional previews, sometimes to fit publication constraints. 230(c)(1) has applied in so many cases fitting that paradigm (People v. Ferrer represents an outer extreme), yet future plaintiffs can argue that any piece excluded from the extract creates a deceptive omission and VOILA! Bye bye 230.

So, for example, Google’s search results descriptions republishes extracts from the source website. This has qualified for Section 230 (e.g., O’Kroley v. Fastcase). This court is saying that 230 would not apply if the plaintiff claims the search results description left behind contextualizing information, which will happen ALL THE TIME. Boom–all of Google search descriptions are now potentially outside Section 230.

The court keeps talking. It says Zeran “cannot be stretched to include actions that go beyond formatting or procedural alterations and change the substance of the content altered.” Yet another false dichotomy. The line between “formatting”/”procedural” changes and “substantive” changes is incoherent. Where do Google search descriptions lifted from third-party indexed sites fit in that spectrum?

Other Stuff

The opinion is filled with many other sharp edges, including:

  • Section 230(c)(1) “is not a license to do whatever one wants online.” Who claims it does? #Strawman.
  • The strawman argument was so nice the judge repeated it twice: Section 230(c)(1) “does not insulate a company from liability for all conduct that happens to be transmitted through the internet.”
  • “Zeran left the door open to finding § 230(c)(1) protection applies when a claim holds a party liable for a decision not to publish, and we need not decide here if we should shut it. Zeran suggested that it might allow § 230(c)(1) to bar claims whenever avoiding liability under those claims would require acting as a publisher. In other words, it is possible to read Zeran as applying § 230(c)(1) protection when an interactive service provider would be held liable for failing to publish information.” This is interesting because a talking point on the “right” has been that decisions not to publish third-party content should be governed exclusively by 230(c)(2)(A) and its good faith standard, not 230(c)(1). This court seemingly concedes the possibility–made explicit by Zeran–that 230(c)(1) applies to “withdrawing” content.
  • “Take a writer of a ransom note who cuts letters out of a magazine to list his demands. That writer might be said to be “altering” content. Yet, the note’s writer is hardly acting as an “editor” of the magazine. Instead, he has substantively changed the magazine’s content and transformed it from benign information about sports or entertainment into threatening information about bags of cash and ultimatums.” Wow, this should have been ditched in the editing. This is like a bad thriller movie, not an enlightening judicial analogy.

Implications

MAGA hates 230 (even though the MAGA social media services depend on 230, and Trump himself personally loves it). Swayed by that animus, Trump’s Federalist Society judges will act as the executioners of 230–no need for Congress to amend 230. This opinion previews the Gonzalez apocalypse, coming to destroy the Internet by June 2023. #MAGA #M230GA.

Case citation: Henderson v. The Source for Public Data LP, 2022 WL 16643916 (4th Cir Nov. 3, 2022)