Investor and Cost Plus Drugs co-founder Mark Cuban says Facebook users are being shown AI deepfake ads that use a fabricated version of his face and voice to sell a diabetes treatment he has never endorsed, and that the people paying for it are overwhelmingly older. Speaking on the Pivot podcast with Kara Swisher and Scott Galloway, Cuban said he hears from those buyers directly, week after week, asking whether the cure they ordered will work.
The accusation surfaced alongside a separate line of reporting from Reuters this November, which described internal Meta documents projecting roughly $16 billion in 2024 revenue — about 10% of the company’s total for the year — from advertising tied to scams and prohibited goods. Meta disputes that figure.
The $294 diabetes cure Cuban never endorsed
Cuban described receiving messages every week from older people asking whether his diabetes cure actually works, or telling him they had already paid $294 for it. He says he never endorsed the product.
“Every single week I get emails from little old men and ladies saying, ‘Will my cure for diabetes work?’ or ‘I just ordered your cure for diabetes for $294.'”
That quote was attributed to Cuban by the International Business Times. The podcast episode itself was not independently verified in available reporting, and the account of Cuban’s remarks rests on secondary coverage from FinanceBuzz and the International Business Times.
According to those reports, the AI deepfake ads relied on AI-generated versions of his face and voice to promote the treatment. Some of the ads reportedly carried fine-print disclaimers noting that the person shown was AI-generated and that the clip was intended for entertainment. Cuban’s position, as summarized in that coverage, is that a disclaimer buried beneath a convincing false endorsement does not undo the damage the endorsement has already done.
Reuters reports a $16 billion internal estimate
Reuters reported this November that internal Meta documents projected approximately $16 billion in 2024 revenue from ads involving scams and prohibited goods, equivalent to roughly 10% of the company’s revenue that year. Reuters characterized the number as an internal company estimate rather than a confirmed accounting figure.
A separate internal estimate reported by Reuters described Meta’s platforms displaying roughly 15 billion higher-risk scam ads per day in December 2024. Internal safety staff also reportedly estimated that Meta played a role in about one-third of scams that successfully affected victims in the United States.
The documents reportedly described a 95% probability threshold that automated systems had to clear before classifying an advertiser as a scammer eligible for banning — a high bar that, as described in the reporting, allows harmful ads to remain live while users continue to buy from them.
Meta calls the documents a selective view
Meta spokesperson Andy Stone rejected the characterization, saying the documents presented “a selective view that distorts Meta’s approach to fraud and scams.” He called the 10% estimate “rough and overly-inclusive” and said complaints about scam ads had fallen 58% over the preceding 18 months.
The company also says it removed more than 134 million scam ads in 2025. That figure comes from Meta itself and has not been independently verified.
Meta’s handling of accounts and advertising on its platforms has drawn scrutiny from other quarters this year as well, including an accusation from the organizing group No Kings that Meta suspended its Instagram account.
FBI figures show older adults bearing the losses
Federal data underscores why the demographic Cuban describes matters. FBI figures show people over 60 reported more than $7.7 billion in losses to online fraud in 2025, which the agency described as 37% higher than in 2024.
Complaints specifically involving AI, across all age groups, accounted for nearly $893 million in reported losses that year.
Those totals are likely to undercount the real scale of the problem. Many victims never realize AI was involved in the pitch they responded to, and many others never report the fraud at all.
AI deepfake ads and the speed of removal
The accountability question raised by the reporting is not whether Meta eventually removes scam ads, but how many people are defrauded by AI deepfake ads in the interval before removal. A disclaimer identifying a video as AI-generated offers a form of disclosure, but it does not protect someone who watched a convincing clip of a recognizable face and ordered a $294 treatment before reaching the fine print.
As the reporting frames it, what the platforms need is earlier detection, faster removal, and enforcement that does not depend on a well-known investor raising the issue publicly to prompt a response. The distance between Meta’s description of its enforcement and what users encounter is measured in billions of dollars of reported losses — and, in Cuban’s account, in the weekly emails he receives from people who trusted a face they recognized in AI deepfake ads.
Full details of the Reuters findings and Cuban’s remarks were compiled in reporting published by Yahoo News.
