The Internet Survives SCOTUS Review (This Time)–Twitter v. Taamneh and Gonzalez v. Google

Today was the 2023 Super Bowl of Internet Law at the U.S. Supreme Court [FN]. SCOTUS issued two eagerly awaited decisions in the Twitter v. Taamneh and Gonzalez v. Google cases (as well as decisions in the Warhol copyright fair use case and the Amgen patent enablement case). Twitter won its decision unanimously, and the Supreme Court per curiam punted the Google case back to the 9th Circuit with the clear message that the plaintiffs should lose. Overall, today was a better-than-expected day for the Internet’s short-term future.

[FN: I say the 2023 Super Bowl because the Supreme Court necessarily will be taking Internet Law cases every term for the foreseeable future, and each new Internet Law case they take has the capacity to rock our world.]

Between the two decisions, we get a powerful opinion on the topic of “aiding and abetting” online, while Section 230 dodged its first SCOTUS review. The rulings should put a decisive end to the genre of lawsuits over social media supporting terrorists; and the Twitter ruling will cast a negative shadow over other cases alleging that social media services facilitate illegal activity. However, the lack of SCOTUS guidance on Section 230 will continue to fuel plaintiffs’ attacks on it, assuming Congress doesn’t outright or functionally repeal it imminently.

A quick reminder about the cases: both involve terrorist attacks where the plaintiffs allege that terrorist organizations used social media to post and disseminate their content. There have been about 20 lawsuits in this specific genre, and there are a growing number of analogous cases, such as the lawsuit filed this week against social media for its alleged contributions to the Buffalo shooting attack.

Twitter, Inc. v. Taamneh, No. 21–1496 (SCOTUS May 18, 2023)

This opinion interprets the JASTA statute, which imposes liability on someone who “aids and abets, by knowingly providing substantial assistance,” which the statute does not define further. The Supreme Court says that the term “aiding and abetting” in the statute should be interpreted using the common law. To do so, the court adopts the “Halberstam” factors [FN] as the common law test for aiding-and-abetting, although the court says factually the Twitter case is a “far cry” from Halberstam, so the court turns “to the common law of aiding and abetting upon which Halberstam rested and to which JASTA’s common-law terminology points.”

[FN: the factors are: (1) “the nature of the act assisted,” (2) the “amount of assistance” provided, (3) whether the defendant was “present at the time” of the principal tort, (4) the defendant’s “relation to the tortious actor,” (5) the “defendant’s state of mind,” and (6) the “duration of the assistance” given. However, the court does not march through them mechanically in this opinion.]

The court then contextualizes some limits on the common-law aiding-and-abetting doctrine. It says:

assume that any assistance of any kind were sufficient to create liability. If that were the case, then anyone who passively watched a robbery could be said to commit aiding and abetting by failing to call the police. Yet, our legal system generally does not impose liability for mere omissions, inactions, or nonfeasance; although inaction can be culpable in the face of some independent duty to act, the law does not impose a generalized duty to rescue…if aiding-and-abetting liability were taken too far, then ordinary merchants could become liable for any misuse of their goods and services, no matter how attenuated their relationship with the wrongdoer.

Applying these principles to this case, the court says that the plaintiffs cannot satisfy the Halberstam factors:

plaintiffs have satisfied Halberstam’s first two elements by alleging both that ISIS committed a wrong and that defendants knew they were playing some sort of role in ISIS’ enterprise. The key question, therefore, is whether defendants gave such knowing and substantial assistance to ISIS that they culpably participated in the Reina attack. The allegations here fall short of that showing under Halberstam’s framework as properly understood by reference to the common-law principles it applied.

Thus, the court accepted the plaintiffs’ allegations that Twitter knew it was “playing some sort of role in ISIS’ enterprise.” That scienter does not carry the day.

The opinion addresses the implication of Twitter’s role as a self-service publication tool that was generally available to all authors. The court says offering such tools does not establish aiding-and-abetting:

the only affirmative “conduct” defendants allegedly undertook was creating their platforms and setting up their algorithms to display content relevant to user inputs and user history. Plaintiffs never allege that, after defendants established their platforms, they gave ISIS any special treatment or words of encouragement. Nor is there reason to think that defendants selected or took any action at all with respect to ISIS’ content (except, perhaps, blocking some of it). Indeed, there is not even reason to think that defendants carefully screened any content before allowing users to upload it onto their platforms. If anything, the opposite is true: By plaintiffs’ own allegations, these platforms appear to transmit most content without inspecting it.

The mere creation of those platforms, however, is not culpable. To be sure, it might be that bad actors like ISIS are able to use platforms like defendants’ for illegal—and sometimes terrible—ends. But the same could be said of cell phones, email, or the internet generally. Yet, we generally do not think that internet or cell service providers incur culpability merely for providing their services to the public writ large.

You can see how plaintiffs might look for opportunities in the negative spaces here. However, start with the court’s baseline: the “mere creation” of self-service publication tools isn’t culpable. The court then contemplates that Twitter would need to make some extraordinary or custom arrangements with ISIS as a content publisher to start showing aiding-and-abetting. Plaintiffs will rarely if ever have the evidence to support this.

But what about the “algorithms”???

plaintiffs assert that defendants’ “recommendation” algorithms go beyond passive aid and constitute active, substantial assistance. We disagree. By plaintiffs’ own telling, their claim is based on defendants’ “provision of the infrastructure which provides material support to ISIS.” Viewed properly, defendants’ “recommendation” algorithms are merely part of that infrastructure. All the content on their platforms is filtered through these algorithms, which allegedly sort the content by information and inputs provided by users and found in the content itself. As presented here, the algorithms appear agnostic as to the nature of the content, matching any content (including ISIS’ content) with any user who is more likely to view that content. The fact that these algorithms matched some ISIS content with some users thus does not convert defendants’ passive assistance into active abetting. Once the platform and sorting-tool algorithms were up and running, defendants at most allegedly stood back and watched; they are not alleged to have taken any further action with respect to ISIS.

What a powerful line this is: “defendants’ ‘recommendation’ algorithms are merely part of that infrastructure.” By definition, self-service publication tools include the ability to post and disseminate content, including delivering the content to audiences that are most likely to be interested. The delivery function is just “part of the infrastructure,” like the rest of the publication offering, and thus does not qualify as a plot to facilitate illegal outcomes. It’s frankly a little shocking to see Justice Thomas come out swinging in favor of algorithms, but here we are.

Plaintiffs can look for soft spots in this passage. For example, they can challenge the distinction between “passive” and “active” recommendations because that distinction is incoherent. By definition, recommendations are never  “passive” or “neutral.” Similarly, the plaintiffs can explore how algorithms can be are “agnostic as to the nature of the content,” which is equally incoherent. Not only do algorithms routinely handle different types of content differently (i.e., photos, text, and videos are all processed differently), but algorithms account for the content’s substance. Finally, the idea that the social media services “stood back and watched” might be a convenient rhetorical device, but the social media services constantly iterate their algorithms in response to a variety of pressures.

Though I wish Justice Thomas had written this passage differently, I’m not particularly worried about any of these soft spots in the language, because the court’s implications are crystal-clear: if the service wasn’t intentionally and manually optimized to help the terrorists, plaintiffs can’t show aiding-and-abetting. And plaintiffs will rarely if ever have that evidence. Given the opinion’s clarity, lower court judges are likely to focus on the opinion’s broad holding, not the less-than-optimal wording details. In the end, even if the court’s characterizations of social media is imprecise, the clear and broad takeaway is that the services as currently configured are not aiding-and-abetting.

The court’s application of its legal standards reinforces how hard it will be for plaintiffs to find a way around this. The court says:

both tort and criminal law have long been leery of imposing aiding-and-abetting liability for mere passive nonfeasance. To show that defendants’ failure to stop ISIS from using these platforms is somehow culpable with respect to the Reina attack, a strong showing of assistance and scienter would thus be required. Plaintiffs have not made that showing.

(“Nonfeasance” is a tautology. The legal standards determine what constitutes malfeasance and nonfeasance).

The court continues:

defendants’ platforms are global in scale and allow hundreds of millions (or billions) of people to upload vast quantities of information on a daily basis. Yet, there are no allegations that defendants treated ISIS any differently from anyone else. Rather, defendants’ relationship with ISIS and its supporters appears to have been the same as their relationship with their billion-plus other users: arm’s length, passive, and largely indifferent.

The services are hardly “indifferent” towards their billion-plus users. The services care a lot about their experiences. And the hypothesized “equal treatment” between ISIS and the non-ISIS users invites plaintiffs to find evidence that the algorithms actually favored ISIS over [pick whatever alternative content the plaintiffs will best help their case]. As with the other details, you have to ignore the bigger picture of the opinion: Twitter provided self-service publication tools to a billion users, and it’s OK that the machines did the rest.

The court continues:

plaintiffs point to no act of encouraging, soliciting, or advising the commission of the Reina attack that would normally support an aiding-and-abetting claim.  Rather, they essentially portray defendants as bystanders, watching passively as ISIS carried out its nefarious schemes.

(Invoking the Seinfeld finale again). To be clear, social media services absolutely encourage, solicit, and advise users to submit content. But the court makes clear that the evidence has to show that they encouraged/solicited/advised with respect to the specific terrorist attack. Plaintiffs will never have this evidence.

The court continues:

plaintiffs identify no duty that would require defendants or other communication-providing services to terminate customers after discovering that the customers were using the service for illicit ends….Even if there were such a duty here, it would not transform defendants’ distant inaction into knowing and substantial assistance that could establish aiding and abetting the Reina attack.

To be clear, the DMCA online safe harbor literally is conditioned on Internet services–even Internet access providers–terminating their relationships with customers who used the service for infringing ends. I’m sure we could find other statutory or common law examples of this.

The court continues:

Given the lack of any concrete nexus between defendants’ services and the Reina attack, plaintiffs’ claims would necessarily hold defendants liable as having aided and abetted each and every ISIS terrorist act committed anywhere in the world. Under plaintiffs’ theory, any U. S. national victimized by an ISIS attack could bring the same claim based on the same services allegedly provided to ISIS. Plaintiffs thus must allege that defendants so systemically and pervasively assisted ISIS that defendants could be said to aid and abet every single ISIS attack….Plaintiffs do not claim that defendants intentionally associated themselves with ISIS’ operations or affirmatively gave aid that would assist each of ISIS’ terrorist acts. Nor have they alleged that defendants and ISIS formed a near-common enterprise of the kind that could establish such broad liability. These allegations are thus a far cry from the type of pervasive, systemic, and culpable assistance to a series of terrorist activities that could be described as aiding and abetting each terrorist act.

The court is 100% correct that the plaintiffs sought to develop a rule of law that would hold Internet services liable for all terrorist attacks anywhere around the globe. It’s great to see the court reject that proposition. However, to be clear, many legislatures are seeking to turn Internet services into financial guarantors against any anti-social behavior online. This opinion does not limit those statutes (though other legal doctrines might), but each one of those regulatory efforts poses an existential threat to the Internet.

Justice Thomas could not resist addressing hypothetical situations in dicta:

There may be, for example, situations where the provider of routine services does so in an unusual way or provides such dangerous wares that selling those goods to a terrorist group could constitute aiding and abetting a foreseeable terror attack….Or, if a platform consciously and selectively chose to promote content provided by a particular terrorist group, perhaps it could be said to have culpably assisted the terrorist group… In this case, it is enough that there is no allegation that the platforms here do more than transmit information by billions of people, most of whom use the platforms for interactions that once took place via mail, on the phone, or in public areas. The fact that some bad actors took advantage of these platforms is insufficient to state a claim that defendants knowingly gave substantial assistance and thereby aided and abetted those wrongdoers’ acts. And that is particularly true because a contrary holding would effectively hold any sort of communication provider liable for any sort of wrongdoing merely for knowing that the wrongdoers were using its services and failing to stop them.

I don’t love this example: “if a platform consciously and selectively chose to promote content provided by a particular terrorist group, perhaps it could be said to have culpably assisted the terrorist group.” Plaintiffs will make all kinds of bogus comparisons to try to show this. I hope courts applying this opinion will recognize the narrowness of this example. Maybe the court is thinking that Musk will decide to embrace ISIS and instruct Ella Irwin to max-boost ISIS content. While we cannot rule out something that crazy from Musk, I assume more rational choices are not what the court intended to cover.

To reinforce the inapplicability of the sea-lioned hypos, the court resumes:

the question is whether defendants gave substantial assistance to ISIS with respect to the Reina attack. The focus thus must remain on the Reina attack; plaintiffs’ failure to allege any definable nexus between the defendants’ assistance and that attack therefore—at minimum—drastically increases their burden to show that defendants somehow consciously and culpably assisted the attack…

The Ninth Circuit thus erred in focusing (as it did) primarily on the value of defendants’ platforms to ISIS, rather than whether defendants culpably associated themselves with ISIS’ actions.

The opinion then emphasizes how it really is trying to set the bar high, such that pretextual plaintiff arguments won’t work:

the Ninth Circuit’s analytic approach thus elided the fundamental question of aiding-and-abetting liability: Did defendants consciously, voluntarily, and culpably participate in or support the relevant wrongdoing? As we have explained above, the answer in this case is no. Plaintiffs allege only that defendants supplied generally available virtual platforms that ISIS made use of, and that defendants failed to stop ISIS despite knowing it was using those platforms. Given the lack of nexus between that assistance and the Reina attack, the lack of any defendant intending to assist ISIS, and the lack of any sort of affirmative and culpable misconduct that would aid ISIS, plaintiffs’ claims fall far short of plausibly alleging that defendants aided and abetted the Reina attack.

Finally, the court turns to the Ninth Circuit’s discussion about YouTube’s revenue-sharing with terrorists who uploaded videos. The Ninth Circuit said those claims were enough to get around Section 230, but they still failed on the prima facie case because the plaintiffs didn’t allege that the amount of revenue or that it was sufficiently substantial. The court agrees:

The complaint here alleges nothing about the amount of money that Google supposedly shared with ISIS, the number of accounts approved for revenue sharing, or the content of the videos that were approved. It thus could be the case that Google approved only one ISIS-related video and shared only $50 with someone affiliated with ISIS; the complaint simply does not say, nor does it give any other reason to view Google’s revenue sharing as substantial assistance. Without more, plaintiffs thus have not plausibly alleged that Google knowingly provided substantial assistance to the Reina attack, let alone (as their theory of liability requires) every single terrorist act committed by ISIS.

Plaintiffs typically will not have the facts the court says are needed when they file their complaint, so their claims should fail for lack of pleading specificity unless courts over-generously let them get into discovery to root for the evidence.

Even so, I was confused about the court’s position here. The court earlier seemed to say that the plaintiffs needed to show that the videos supported the specific terrorist attack, not just the organization generally. So it seems like the substantial financial assistance would somehow need to link the video and the revenue to the attack? The court doesn’t make this clear, but it seems like the plaintiffs won’t be making much progress if they just show a more substantial revenue share amount that isn’t attack-specific.

The court summarizes its conclusion:

both the common law and Halberstam provide some clear guideposts: The point of aiding and abetting is to impose liability on those who consciously and culpably participated in the tort at issue. The focus must remain on assistance to the tort for which plaintiffs seek to impose liability. When there is a direct nexus between the defendant’s acts and the tort, courts may more easily infer such culpable assistance. But, the more attenuated the nexus, the more courts should demand that plaintiffs show culpable participation through intentional aid that substantially furthered the tort. And, if a plaintiff ’s theory would hold a defendant liable for all the torts of an enterprise, then a showing of pervasive and systemic aid is required to ensure that defendants actually aided and abetted each tort of that enterprise.

Here, however, the nexus between defendants and the Reina attack is far removed. As alleged by plaintiffs, defendants designed virtual platforms and knowingly failed to do “enough” to remove ISIS affiliated users and ISIS-related content—out of hundreds of millions of users worldwide and an immense ocean of content—from their platforms. Yet, plaintiffs have failed to allege that defendants intentionally provided any substantial aid to the Reina attack or otherwise consciously participated in the Reina attack—much less that defendants so pervasively and systemically assisted ISIS as to render them liable for every ISIS attack.

I anticipate adding this opinion to my Internet Law casebook, though I have to think about where it fits because Section 230 covers so many of the potential claims where common law aiding-and-abetting would otherwise come up.

Gonzalez v. Google LLC, No. 21–1333 (SCOTUS May 18, 2023

[A reminder that I filed an amicus brief in this case in support of Google, explaining how Section 230 intersects with the First Amendment.]

The court says the Taamneh ruling shows that the plaintiffs are likely to lose the Gonzalez case on its merits, so there’s no point in discussing Section 230 because its application is essentially hypothetical:

it has become clear that plaintiffs’ complaint—independent of §230—states little if any claim for relief. …we think it sufficient to acknowledge that much (if not all) of plaintiffs’ complaint seems to fail under either our decision in Twitter or the Ninth Circuit’s unchallenged holdings below. We therefore decline to address the application of §230 to a complaint that appears to state little, if any, plausible claim for relief.

It would have been great to get a pro-230 ruling from the Supreme Court, which seemed likely based on the tenor of oral arguments. However, I worried that the opinion would contain qualifications or hypotheticals (like the ones I critiqued above in the Taamneh case) that would spur plaintiffs even if the overall opinion was defense-favorable. So I’m glad we avoided that. Yet, by not providing any input on Section 230, the Supreme Court has left room for all of the plaintiffs to keep pushing their anti-230 arguments in the lower courts. The Supreme Court will have to resolve this eventually.


  • I’m sympathetic to the plaintiffs. They are victims of terrorist attacks. However, unfortunately they chose to sue the wrong defendants. It’s heartbreaking to see the courts tell them “no,” but that has always been the correct outcome.
  • As I mentioned, there were about 20 cases in the genre of social media services providing “material support for terrorists.” Those cases were poorly framed and litigated at every step along the way. They were never meritorious–and they had all uniformly failed before the Ninth Circuit’s Gonzalez ruling. That’s why the Ninth Circuit’s Gonzalez ruling was so shocking. It found merit in cases that everyone else had rejected.
  • The Ninth Circuit was wrong, of course, and this unanimous Supreme Court opinion makes that clear. The court says: “the Ninth Circuit went astray through a series of missteps that, together, obscured the essence of aiding-and-abetting liability.” In other words, not even close. The fact that the Ninth Circuit didn’t rehear this case en banc is an indictment of the entire circuit. They should have cleaned their house before SCOTUS got involved.
  • This case only involved the JASTA statute. However, because the Supreme Court said that JASTA built upon common law principles, the Supreme Court’s discussion should apply to all situations where common law aiding-and-abetting is at issue. That makes this a helpful ruling for other Internet Law cases; and the ruling should apply well outside the Internet Law context. Justice Jackson’s concurrence claims the holding is narrow, but she was just one vote.
  • This opinion only applies to common law aiding-and-abetting, so it leaves undisturbed many of the other Internet Law battlefronts, including the voluminous number of new anti-Internet statutes being enacted by the states as well as the various 230 workaround claims in vogue with the plaintiffs, such as products liability claims. I expect defendants will find many good passages to cite in those other legal battles, but only as persuasive authority. Similarly, even if plaintiffs completely lose the ability to assert common law aiding-and-abetting, they have a rich toolkit of claims to bring–and they may find some of the soft spots in this opinion as helpful persuasive authority for those claims.
  • This opinion reiterates, for the nth time, that Section 230 is not the only reason why defendants win cases involving liability for third-party content. This time, the plaintiffs failed the elements of the prima facie case, though the cases have also failed on other grounds, like the First Amendment and causation. It makes me question the good faith of anyone who cites the Taamneh case, or any of the material-support-for-terrorist cases, as reasons to amend Section 230.
  • This prompts my frequent reminder that changing Section 230 would not change the results of this case. Instead, to the extent that Section 230 is a procedural fast-lane to reach the same substantive results as the other applicable doctrines, amending Section 230 would make everyone worse off.
  • It’s tempting to engage in the cynical speculation that Justice Thomas–knowing that other Internet Law cases are coming to the court soon–is playing 4D chess in this opinion. For example, could this opinion lay the foundation for a future SCOTUS evisceration of Section 230, on the basis that the Internet services shouldn’t be too upset because they will have other common law defenses?
  • Most of the amicus energy went into the Gonzalez case, but it turns out the real action was the Twitter case. It’s a reminder that every Internet Law case that reaches the Supreme Court has the potential to be a big deal!
  • Even though the Internet’s status quo survived today, there are many important cases hurtling towards the Supreme Court, including the pending NetChoice cases, constitutional challenges to the age verification statutes, constitutional challenges to the Montana TikTok ban, etc., etc. Each of those cases pose strategic importance to the Internet’s future. The legislatures, especially at the states, are in the throes of a full-blown techlash, and only the courts can save us from their poor policy decisions. The Internet has to win each and every one of those future cases just to preserve the status quo. The chances of running the table for the indefinite future feel vanishingly small, even as we enjoy today’s pro-Internet rulings.

I want to give a shoutout to the plaintiffs’ lawyers in these cases for their important contributions to Internet law. It may not be the law they wanted to make, but they can take some pride in knowing that they helped an often-fractured court find unanimity.