# Section 230 Doesn’t Apply to Generative AI Enhancements to Ad Copy (But the Plaintiffs Lose Anyway)–Bouck and Suddeth v. Meta > Source: > Published: 2026-06-13 16:51:34+00:00 # Section 230 Doesn’t Apply to Generative AI Enhancements to Ad Copy (But the Plaintiffs Lose Anyway)–Bouck and Suddeth v. Meta [The blog post covers two cases involving scammy ads on Facebook that were part of a pump-and-dump for Chinese penny stocks.](https://blog.ericgoldman.org/wp-content/uploads/2026/02/IMG_2022.jpg) The first two rulings came in March. In the Bouck case, the court rejected Facebook’s Section 230 defense because Facebook’s generative AI allegedly contributed to the ad copy. In the Suddeth case, the court accepted Facebook’s Section 230 defense because the plaintiffs focused on algorithmic amplification. Although the Bouck case initially overcame Section 230, in an under-the-radar ruling this week, the court nevertheless dismissed it as preempted by federal securities law. So the Bouck plaintiffs got a first-hand taste of the infamous Ninth Circuit switcheroo, which occurs when the plaintiffs get false hope that they might actually win because they got around Section 230, only to slam into other brick walls in their prima facie case. **The March Bouck Ruling** *Section 230* Meta’s Section 230 defense turns on whether it was a co-creator of the ads sufficient to become an “information content provider.” The court says “What it means to “create” or “develop” content on the internet is not self-evident.” The court says there’s a fact dispute over whether Facebook’s contributions were material: The alleged illegality stems from the advertisements’ content—i.e., the false statements made to Facebook and Instagram users that induced them to click on the ads. Plaintiffs have averred that Meta participated in the construction of the ads by literally generating, using artificial intelligence, the images and text in the advertisements. That degree of participation is not protected by section 230…. [cite to [Forrest v. Meta]]The district court in Forrest accepted that optimizing the appearance of an ad to drive engagement was enough of a contribution to the ads’ illegality to preclude section 230 immunity. Here, in addition to averring facts which, if proven, would establish that Meta altered the ads’ appearance to maximize impressions, Plaintiffs have averred that Meta’s tools allowed the scammers to produce “AI-generated text and images” for use in the ads through its Advantage+ Creative tool. The court says Carafano doesn’t help Facebook because: Plaintiffs have averred that Meta created the offending information by generating some of the false statements that tricked them into the investment scheme…. Plaintiffs aver that the scammers used Meta’s Advantage+ Creative tool which, as explained, uses artificial intelligence to enhance whatever message the user inputs. If a user, for example, tells the tool that he is interested in an ad promising astronomical weekly investment returns, Advantage+ Creative will spin up a slew of ads that include the provided language and other language, images, and videos it decides will be effective in promoting the user’s chosen message…. Without question, Advantage+ Creative and the other tools in Meta’s advertising suite would not have come up with that language without the inspiration from the scammers, but that language is still the creation of Meta. One way of reading this decision is that Section 230 has limited applicability to Generative AI outputs. If the model outputs something new (as opposed to verbatim replicating material in its index or provided by the user), then the newly created material isn’t covered by Section 230. *Aiding and Abetting Fraud* [The court says Facebook’s ad review process should have detected that the ad looked scammy (see an example on the right): “Even a cursory look would warrant suspicion that the ad is fraudulent. Meta cannot, with a straight face, claim otherwise.” Facebook responded that the ad review was automated (i.e., no human performed the “cursory look” that the judge was contemplating), a response the judge calls “confounding” because “It was Meta’s decision to use technological review tools to screen ads, and it does not now get to claim it had no idea what was going on because it tasked some software program with doing the first pass.”](https://blog.ericgoldman.org/wp-content/uploads/2026/06/bouck.jpg) The judge is dabbling with some heady topics here in an unsatisfying and superficial way. At core, the judge–whether he intended to or not–is addressing the epistemological question of when a machine “knows” something. This is a crucial topic for the digital age, and it deserves more in-depth and thoughtful treatment than the judge provides here. Alternatively, the judge is accepting an argument that it’s “willful blindness” to turn over ad review to the machines. But nowadays machines do a lot of scanning and screening without humans in the loop, and it deserves some careful and thoughtful judicial review to determine if such delegation deserves to be condemned with a “willful blindness” style punishment. The judge didn’t do that either. *Contract Breach* The plaintiffs tried the oh-so-tired hack of claiming that TOS content policy restrictions should be treated as affirmative representations that the policies won’t be violated. Not this again. Sigh. The judge doesn’t take the bait (cite to [Lloyd v. Facebook](https://blog.ericgoldman.org/archives/2024/12/facebook-defeats-users-tos-breach-claim-lloyd-v-facebook.htm)): The provision of the ToS on which Plaintiffs rely does not expressly or impliedly impose a binding contractual obligation on Meta to do anything. It is much more naturally read as a creating a duty of its users not to pollute Meta’s platforms with scam investment ads…. To the extent the ToS even mentions Meta doing something to prevent fraud, it speaks only in aspirational terms…Meta, however, never promises to take concrete steps to effectuate that aspiration. *Negligence* “Plaintiffs have averred that Meta did more than just sit idle as fraudsters roamed freely on their platforms. Therefore, no “special relationship” need be pleaded for the case to move forward.” *Unruh Act* “Plaintiffs here aver that they were targeted because of their race or national origin, not that they were excluded from anything. Whatever moral condemnation that merits, it is not a violation of the Unruh Act.” The court distinguishes [Liapes](https://blog.ericgoldman.org/archives/2023/10/does-californias-anti-discrimination-law-ban-ad-targeting-liapes-v-facebook.htm) because, in that case, the plaintiff complained she didn’t receive ads based on her protected classifications. Here, the plaintiffs got the ads: “Far from encountering an exclusionary practice, they encountered an inclusionary one—it is just that they wish they were not included.” The court rejects the plaintiffs’ attempt “to spin Liapes into a general prohibition on targeting based on protected characteristics.” *Case* *Citation*: [Bouck v. Meta Platforms, Inc.](https://storage.courtlistener.com/recap/gov.uscourts.cand.451567/gov.uscourts.cand.451567.61.0.pdf), 2026 WL 810036 (N.D. Cal. March 24, 2026) **The March Suddeth Ruling** Because of its implications for Generative AI, the Bouck case has garnered some coverage. That coverage overshadowed a companion case, the Suddeth decision, issued by the same judge, on the same day, involving the same basic claim (Chinese stock pumping-and-dumping). Unlike the Bouck case, the judge dismissed the Suddeth case. In Bouck, the plaintiffs claimed that Facebook helped the advertiser build and polish up the ads using Generative AI. In Suddeth, the plaintiffs claimed Facebook algorithmically amplified the ads. The court has little difficulty concluding that algorithmic amplification is governed by Section 230, citing [Dyroff](https://blog.ericgoldman.org/archives/2019/08/a-significant-section-230-defense-win-in-the-ninth-circuit-dyroff-v-ultimate-software.htm) and [Doe v. Grindr](https://blog.ericgoldman.org/archives/2025/02/ninth-circuit-says-section-230-preempts-defective-design-claims-doe-v-grindr.htm): Their theory of algorithmic amplification is nothing more than an averment of facilitation. The core illegality—the choice to use Plaintiffs’ likeness and falsely represent that they endorsed certain investments—was exclusively undertaken by the scammers. Meta provided those scammers tools to disseminate that fraud that may well have played a role in the success of the ploy. However, as in both Dyroff and Grindr, Meta’s tools were content neutral on their own—it was the scammers who chose to manipulate those tools for illicit ends. The judge distinguishes his own simultaneous ruling in Bouck: [in Bouck,] the plaintiffs averred that Meta contributed materially to the development of the ads by offering generative-AI tools that developed the ultimate content of the fraudulent ads. Meta’s role in that scheme, therefore, allegedly went beyond offering neutral tools that promoted content developed exclusively by the scammers—Meta, at least according to the complaint, was a genuine co-conspirator in the creation of the offending content. Plaintiffs have failed to aver a similar level of complicity here. The court summarizes: “Section 230 thus bars any claim which, at bottom, seeks to hold Meta liable for the damage done by the content of the fraudulent ads.” *Case Citation: *[Suddeth v. Meta Platforms, Inc.](https://storage.courtlistener.com/recap/gov.uscourts.cand.457604/gov.uscourts.cand.457604.41.0.pdf), 2026 WL 810252 (N.D. Cal. March 24, 2026) **The June Bouck Ruling** We’re back to the classic Ninth Circuit switcheroo: the arguments that the plaintiffs used to get around Section 230 ensure the failure of the prima facie case. Here, the plaintiffs alleged state law claims to redress what is fundamentally a federal securities law claim. The court summarizes: Meta…argues that the theory which helped Plaintiffs defeat the first motion to dismiss compels granting the second. If it is true that Meta contributed to the creation of the fraudulent ads, then this suit is necessarily based on the falsity of Meta’s statements. A suit in which a plaintiff claims the defendant made false statements which led the plaintiff to purchase securities when he otherwise would not have is quintessentially one sounding in the securities laws, even if the right of action comes from state law. SLUSA prevents precisely that type of suit from being maintained in any court, state or federal…. At bottom, Plaintiffs are trying to have it both ways. They assert Meta’s misrepresentations aided and abetted the core fraud by pushing them into scam investment groups while simultaneously maintaining that those misrepresentations were not material to their decision to purchase CLEU stock. Both cannot be true—either the misrepresentations mattered (in which case SLUSA applies) or they did not (in which case their claims fail on the merits). This denouement will surely attract less attention than the March ruling. The plaintiff could in theory overcome this ruling by bringing a federal securities act claim. However, I suspect the plaintiffs won’t due to the significant pleading challenges. Plus, it will be difficult or impossible to put Facebook on the hook for those claims. *Case Citation*: [Bouck v. Meta Platforms Inc.](https://storage.courtlistener.com/recap/gov.uscourts.cand.451567/gov.uscourts.cand.451567.83.0.pdf), 2026 WL 1697630 (N.D. Cal. June 11, 2026)