Cuban: Facebook knowingly runs AI deepfake scam ads; docs say $16B fraud revenue
Mark Cuban says scammers are using an AI copy of his face and voice in Facebook ads for a diabetes treatment he never endorsed. Yahoo Finance reports that he calls the ads deepfakes that Meta runs anyway, and that some include fine print describing the on-screen figure as AI-generated and meant as entertainment. IBTimes reports that Cuban said on a podcast in August that he receives emails every week from older people who believe he sells the treatment.
The $16 billion figure comes from internal Meta documents that Reuters reviewed and reported in November 2025. Meta projected that about 10% of its 2024 revenue, roughly $16 billion, would come from ads for scams and banned products. A December 2024 document estimated that users across Meta's apps saw about 15 billion higher-risk scam ads a day, and a late 2024 document valued that category at about $7 billion a year. Under the documents, Meta's automated systems must judge an advertiser at least 95% likely to be a scammer before banning it.
Meta disputes the framing. Spokesman Andy Stone told Reuters the documents present a selective view and called the 10% estimate rough and overly-inclusive, according to Yahoo Finance. He said complaints about scam ads fell 58% over the prior 18 months and that Meta removed more than 134 million scam ads in 2025.
Why it matters
For AI companies, the case shows that ad review can become a liability question when the ad itself is a synthetic copy of a real person. Voice and face cloning tools are cheap to build and easy to misuse, so startups in generative video, audio or avatar products should expect buyers, regulators and insurers to ask how they verify consent and likeness. A disclaimer line, like the fine print Cuban described, did not keep the ads from reaching older users.
For investors, the documents matter because they show that revenue from flagged ads was weighed against possible fines. IBTimes reports that the internal projections put potential penalties at up to $1 billion, well below the revenue from the higher-risk category. Diligence on consumer-facing AI businesses should therefore cover how they price risky traffic, what fraud thresholds they use, and whether leadership has tied enforcement to revenue targets.
Key points
- 1Cuban says AI copies of his face and voice were used in Facebook ads for an unendorsed diabetes treatment.
- 2Internal Meta documents projected about $16 billion, or 10% of 2024 revenue, from scam and banned-product ads.
- 3Meta's automated systems need at least 95% confidence that an advertiser is a scammer before banning the account.
- 4Meta called the 10% estimate rough and overly-inclusive and said scam-ad user complaints fell 58% over 18 months.
- 5
