42 State AGs Tell Meta: Your AI-Generated Investment Scams Are Your Problem

The Letter When 42 state attorneys general sign the same letter, it's not a suggestion. The coalition, led by New York's Letitia James, gave Meta a clear ultimatum: implement meaningful preventive measures against AI-generated investment scams — or stop running investment ads on Facebook and Instagram altogether. The letter described a fraud machine built on Meta's own products: The Scam Machine The complaint lays out a multi-stage pipeline that weaponizes every piece of the AI stack: AI-generated ads. Scammers use generative AI to create fraudulent investment and "pump-and-dump" ads on Facebook and Instagram, featuring deepfakes of trusted public figures — most notoriously Warren Buffett — endorsing schemes that don't exist. Platform algorithms amplify them. The ads are pushed to precisely the audiences most likely to engage: older users, novice investors, and people already searching for get-rich-quick opportunities. Off-platform follow-through. Victims who click are herded into WhatsApp groups, where the scam continues off Meta's own ad review radar. AI voice-masking. Once trust is established, scammers use voice-cloning tools to impersonate brokers, advisors, or even family members on calls — the same technique behind the AI-enhanced phishing campaigns we've documented — extracting payments that victims believe are legitimate investments. The result, per the AGs: hundreds of millions of dollars stolen from consumers, with each dollar laundered through a chain of platforms that each claim the fraud was someone else's problem. What the AGs Want Specifically, the coalition demanded Meta: Screen advertisers meaningfully — verify who's buying ads, not just process payment. Pre-screen investment ads for fraud indicators before they run, instead of relying on post-hoc takedowns. Use its own AI tools defensively — Meta can detect deepfakes; the AGs want it to, at scale, on its own platforms. Report back, or face enforcement action. Why Platforms Keep Allowing It The economics explain the inertia: fraudulent ads are revenue. Every deepfake-Buffett ad that runs pays Meta's ad network before the inevitable takedown. The takedown-and-refund cycle is cheaper than pre-screening — which is exactly why the AGs are threatening the blunt instrument of banning investment ads entirely, the way the EU moved to criminalize the deepfakes themselves. This is the same pattern we've tracked in celebrity deepfake crypto scams, CEO impersonation attacks, and the democratization of deepfake fraud tools: the AI that generates the scam is the same AI that could detect it — and platforms only deploy it when regulators force the math to change. What to Do Treat investment ads like spam. Legitimate wealth is not built through ads for celebrity-endorsed crypto schemes. Verify before you trust. If a "Buffett" or "Musk" is telling you to send money, it's a deepfake. Hang up, close the tab, and verify through the person's real, official channels. Never move money to "brokers" you met in a chat group. WhatsApp groups, Telegram channels, and Discord servers are where scams finish the job. Report the ads. Takedowns are slow, but every report adds pressure — and the AGs' letter is exactly the kind of pressure that converts reports into enforcement. The Pattern The June 2025 coalition letter was a turning point: it marked the first time a large bloc of state enforcers explicitly named AI-generated fraud as a platform liability, not a user education problem. The era of "we're just a platform" is ending for ad networks — and the AI-scam ecosystem is going to have to find somewhere else to run.