Bouck v. Meta — SLUSA Preempts AI Investment Fraud Class Action After Section 230 Defense Falls

Background

This is the second act of a high-stakes lawsuit over AI-generated investment fraud. A group of plaintiffs lost more than $300 million after falling victim to a sophisticated pump-and-dump scheme that ran through Meta’s platforms. Scammers placed ads on Facebook and Instagram — featuring celebrity likenesses such as Kevin O’Leary and Savita Subramanian (Bank of America’s head of U.S. equity strategy) — funneling victims into private WhatsApp groups where fraudsters impersonating financial advisors pushed them to buy shares of China Liberal Education Holdings Ltd. (CLEU), a near-worthless Chinese penny stock. CLEU shares briefly rose from $5.32 to $7.90, then cratered to $0.15 when the scheme collapsed in late January 2025.

Plaintiffs sued Meta — not the scammers — arguing that Meta’s own artificial intelligence tools, specifically its Advantage+ Creative feature that uses generative AI to generate and modify advertising text, images, and video, made Meta a co-creator of the fraudulent content. That stripped Meta of Section 230 immunity, which normally shields platforms from liability for third-party content. In a March 24, 2026 ruling (previously summarized on LexSummary), the court agreed and denied Meta’s first motion to dismiss, holding that a platform that “literally generat[es], using artificial intelligence, the images and text in the advertisements” cannot claim Section 230 protection.

Having survived Section 230, plaintiffs expected to proceed to discovery. Instead, Meta filed a second motion to dismiss on a completely different jurisdictional ground: the Securities Litigation Uniform Standards Act of 1998 (SLUSA).

The Court’s Holding

Chief Judge Richard Seeborg granted Meta’s motion and dismissed the case, without prejudice, for lack of subject-matter jurisdiction.

SLUSA bars any “covered class action” based on state law claims alleging misrepresentations or deceptive devices “in connection with the purchase or sale of a covered security.” The central question — drawn from the Supreme Court’s decisions in Merrill Lynch v. Dabit (2006) and Chadbourne & Parke v. Troice (2014) — was whether the fraudulent statements made in Meta’s AI-generated ads were “material to” plaintiffs’ decisions to buy CLEU stock.

The court found they plainly were. Even though the ads never mentioned CLEU by name, they gave the scam WhatsApp investment groups an “air of legitimacy” by falsely associating them with respected financial figures. Under plaintiffs’ own theory, absent those misrepresentations they would never have joined the WhatsApp groups and would never have purchased CLEU shares. The alleged fraud was therefore material to — and inextricably connected to — a covered securities transaction. SLUSA preemption followed as a matter of law.

The ruling exposes a striking irony. Plaintiffs won Round One by arguing that Meta’s generative AI tools made Meta a co-creator of the fraudulent content — which defeated Section 230. But that same theory sprung the SLUSA trap in Round Two: if Meta actively co-created the misleading ads, this is a case about deceptive statements that caused securities purchases, which is exactly what SLUSA preempts. As the court put it, plaintiffs were “trying to have it both ways” — asserting that the misrepresentations drove them into scam investment groups, while simultaneously arguing those misrepresentations were not material to their decision to buy CLEU stock. “Both cannot be true,” the court wrote.

The dismissal is without prejudice. Plaintiffs may refile under federal securities law, though they would then face the heightened pleading requirements of the Private Securities Litigation Reform Act of 1995 (PSLRA) — a substantially steeper barrier than the state law claims they originally brought.

Key Takeaways

• Platforms that use generative AI to modify or create advertising content may lose Section 230 immunity — but in investment-fraud cases, plaintiffs who win that argument will immediately face a SLUSA challenge that is difficult to escape.
• SLUSA’s “in connection with” requirement does not require the misstatement to mention a specific security. It is enough that the fraudulent statement was material to plaintiffs’ decision to purchase covered securities, even indirectly.
• The Section 230 / SLUSA catch-22: the theory that defeats platform immunity (AI co-creates the content) is the same theory that triggers federal preemption of state law class actions (the content caused securities transactions).
• The dismissal is without prejudice. Plaintiffs may attempt to bring federal securities claims under PSLRA, but face heightened pleading standards and potential loss of certain state law remedies.
• For fraud schemes that route victims into securities transactions, platform defendants now have a potent two-step defense toolkit: Section 230 in the first instance, and SLUSA as a fallback if Section 230 fails on AI-generation grounds.

Why It Matters

The Bouck saga is shaping up to be a landmark in the law of AI platform liability. The March 24 ruling was already significant for holding that generative AI advertising tools can strip a platform of its Section 230 shield when the AI materially creates the fraudulent content. The June 11 ruling reveals the practical limits of that breakthrough: even when Section 230 falls, federal preemption can still shut the courthouse door — at least for fraud schemes that culminate in securities transactions.

For the millions of Americans victimized by AI-powered investment fraud on social media, this ruling illustrates how difficult it is to hold platforms accountable through state court litigation. The case also sets up a future question: if plaintiffs refile under federal securities law, can they plead around the PSLRA’s demanding requirements? And more broadly, as generative AI becomes embedded throughout social media advertising, courts will increasingly need to decide where the line falls between a platform that “hosts” third-party content and one that “creates” it — and what liability follows from that distinction.

Surfaced via Eric Goldman’s Technology & Marketing Law Blog.