Meta documents point to $16 billion in scam-linked ad revenue as Cuban blasts deepfake ads

Mark Cuban says Facebook keeps running scam advertisements built from an AI replica of his face and voice, and that the company does so while fully aware the viewers being targeted are losing money. According to Yahoo Finance, the investor laid out the allegations on an August episode of Pivot, the podcast hosted by Kara Swisher and Scott Galloway, and described Facebook as the worst platform he knows of on this front.

Deepfake ads wearing Cuban's face

Cuban says that nearly every week he hears from elderly customers asking whether a $294 diabetes cure they just ordered will actually work. He has never sold any such product. The ads promoting it are deepfakes assembled from AI copies of his face and voice, and in his telling Meta approves and runs them anyway, knowing that vulnerable people are being cheated.

The detail matters partly because Cuban co-founded Cost Plus Drugs, an online pharmacy selling inexpensive generic medications. A fabricated medical endorsement in his name is unusually convincing bait.

Some of the ads carry a small disclaimer noting that the person on screen is AI-generated and that the clip is meant as entertainment. Cuban argued the warning is worthless in practice, because by the time anyone reads it the sales pitch has already landed. His broader point was about culture: a company willing to let scammers go after people's grandparents, he suggested, has revealed exactly what it will tolerate everywhere else.

What Meta's own documents say

Cuban's account lines up with reporting on material produced inside Meta itself. On Nov. 6, 2025, Reuters reported on internal Meta documents indicating the company expected roughly $16 billion — about 10 percent of its 2024 revenue — to come from ads pushing scams or prohibited products.

By Meta's internal count, users across its apps were seeing some 15 billion "higher risk" ads per day, a category covering ads with obvious markers of fraud. A late 2024 document valued that category alone at around $7 billion a year. A separate May 2025 presentation by Meta's own safety staff estimated that the company's platforms play a part in roughly a third of the scams that succeed in the United States.

The 95 percent rule

Perhaps the most consequential detail concerns enforcement. The documents indicate that before Meta bans an advertiser, its automated systems must judge the account at least 95 percent likely to be running a scam. Below that threshold, a suspected scammer is not removed. Instead, the advertiser pays more per ad and stays live on the platform.

Meta disputes the framing

Meta spokesman Andy Stone told Reuters the documents offer a selective picture that misrepresents the company's handling of fraud and scams, and he described the 10 percent estimate as rough and overly broad. He also offered counterpoints: user complaints about scam ads fell 58 percent over the prior 18 months, and Meta removed more than 134 million scam ads in 2025. Reuters noted that the same document set reflects a Securities and Exchange Commission inquiry into Meta over advertising tied to financial fraud.

Why it matters

Generative AI has pushed the cost of a convincing impersonation toward zero, and ad platforms are the distribution channel that converts those impersonations into revenue. The reporting raises a structural question: when a meaningful slice of ad income is linked to fraud and enforcement only triggers above a high confidence bar, the economic incentive is to keep borderline advertisers paying. For users, that means deepfakes of trusted figures selling fake cures. For Meta, it means regulators, including the SEC, asking who profited and how much. The distance between Cuban's weekly inbox and Meta's removal statistics is exactly where the platform-accountability debate now sits.