Quick Answer: Facebook deepfake scam ads use AI copies of real people's faces and voices to sell fake cures and products. Mark Cuban says Meta runs them knowingly, and Meta disputes that framing. Reuters reported last November that Meta's internal documents projected about $16 billion, roughly 10% of 2024 revenue, from scam and prohibited-product ads.
Mark Cuban keeps getting emails from people asking whether his $294 diabetes cure works. He has never sold one. The ads, he said on an episode of the podcast Pivot, are Facebook deepfake scam ads built from an AI copy of his face and voice, and Meta runs them anyway. That's a serious accusation from a billionaire investor. The more useful story is in Meta's own documents, which put numbers on how much fraud the platform carries.
What Cuban Is Actually Claiming
Cuban's argument is blunt. He says the ads use his likeness without permission, that some carry a small disclaimer saying the person on screen is AI-generated and the clip is entertainment, and that Meta lets them run. "Facebook knows the little old ladies are getting ripped off," he said, calling the platform "the worst of the worst."
Cuban runs Cost Plus Drugs, an online pharmacy, so he has a direct stake in how people buy medications. That doesn't make him wrong, but it's worth knowing when you weigh his view. Meta hasn't confirmed his account, and Moneywise reported it didn't get an immediate response to his comments.
Here's the part most readers miss. Cuban's claim is about intent: that Meta knows. The documents don't prove intent in the legal sense, but they show the company tracked the problem closely and made revenue decisions around it.
What Meta's Internal Documents Show
Reuters reported on internal Meta documents from November 2025. The key figures, as reported, were:
- Meta expected roughly $16 billion, about 10% of its 2024 revenue, to come from ads pushing scams and prohibited products.
- Users across its apps saw about 15 billion "higher risk" ads a day, meaning ads with obvious fraud markers, by Meta's own internal count.
- A late 2024 document valued that higher-risk category at about $7 billion a year.
- A May 2025 presentation from Meta's safety staff estimated the company's platforms play a part in a third of the scams that succeed in the U.S.
Meta's spokesman, Andy Stone, told Reuters the documents "present a selective view that distorts Meta's approach to fraud and scams." He called the 10% estimate "rough and overly-inclusive." He also said complaints about scam ads fell 58% over the prior 18 months and that Meta removed more than 134 million scam ads in 2025.
Both sides have a point. An internal projection isn't the same as final revenue, and "overly-inclusive" may be a fair criticism of how the category was drawn. Still, a company's own staff estimating a third of successful U.S. scams involve its platforms is hard to wave away. The Securities and Exchange Commission is also reportedly looking into Meta over advertising tied to financial fraud.
The 95% Rule Most Coverage Skips
This is the counter-intuitive part, and it changes how you should think about the problem.
The common advice goes like this: report scam ads and the platform takes them down. The documents suggest the system works differently. Before Meta bans an advertiser, its automated systems must judge the account at least 95% likely to be running a scam. Below that threshold, the suspected scammer isn't removed. The advertiser pays more per ad and keeps running.
Here's what that means in practice. When a platform's enforcement threshold is a probability, not a certainty, borderline advertisers become a pricing question. A scammer who clears the bar can pay higher auction prices and still get served. The platform collects more per impression from the people most likely to be defrauded. Meta's documents, as reported, describe exactly this trade-off.
That doesn't prove Meta chose this on purpose. Every fraud filter has to balance false positives against false negatives. Banning legitimate small businesses for looking suspicious is a real cost, and no platform wants to explain that to its advertisers. But the trade-off has a price, and it falls on users. Note that Meta says its approach is being misrepresented, and that dispute is worth weighing alongside the numbers.
Why Older Adults Pay the Heaviest Price
The FBI's Internet Crime Complaint Center (IC3) 2025 Internet Crime Report puts the human cost in numbers. People 60 and older reported losing $7.7 billion across 201,266 complaints in 2025, which the bureau described as 59% more money than the year before.
The breakdown matters:
- Investment schemes caused the most damage, at about $3.5 billion.
- Tech and customer support scams took another $1.04 billion.
- AI appeared in more than 3,100 complaints, and those victims lost over $352 million.
The FBI also froze $32.9 million tied to older victims' cases in 2025. That's a real recovery, but the agency can only chase money it knows about. Most scam victims never file a report, especially when they're embarrassed or don't know where to report. The real losses are almost certainly higher than these numbers, and the gap is worst among people who feel too ashamed to tell anyone.
Here's where it gets uncomfortable. Deepfake ads don't need to fool everyone. They need to fool a small percentage of a very large audience. A pitch that converts one viewer in a thousand can still produce a steady stream of victims.
Why the Disclaimer Doesn't Protect Anyone
Cuban says some of the ads include a line admitting the person is AI-generated and the clip is entertainment. On paper, that's transparency. In practice, it rarely works as a warning.
Think about how a feed works. You see a familiar face, hear a familiar voice, and read a price that sounds like a deal. The disclaimer, if you see it at all, shows up after the emotional hook has already landed. Cuban put it this way: "By the time anyone gets to it, the pitch has already worked."
That's the cause-and-effect chain that matters. A label placed where few people read it protects the advertiser's record more than the viewer's wallet. Labels still matter in some contexts, and they're better than nothing. But treating a disclaimer as proof an ad is safe is one of the most expensive mistakes a reader can make.
How to Check a Facebook Ad for Scams
You don't need to wait for Meta to fix its enforcement threshold. Here's a practical routine that takes about two minutes:
- Treat any famous face selling a miracle product as fake until proven otherwise. Cuban included. Celebrity endorsements for cures, crypto, and "guaranteed" investment returns are a classic pattern.
- Search the product name with "scam" next to it. Real businesses usually have a track record of reviews, complaints, and press coverage. Scam products often have almost none, or a pile of warnings.
- Check the advertiser through Meta's public Ad Library. It lets you see the ads a page is running, which helps you spot the same creative repeated across many accounts.
- Warn the family member who uses Facebook most. Older relatives are the usual targets, so one direct message can do more than any platform policy.
- Report it the same day at ic3.gov. Fast reports help investigators follow the money before it moves.
Why the same-day rule? Scam payments are often routed through hard-to-reverse channels like gift cards, crypto transfers, or wire payments. The first few hours matter more than the first few weeks.
Frequently Asked Questions
What is a Facebook deepfake scam ad?
A Facebook deepfake scam ad uses AI-generated video or audio to imitate a real person, often a celebrity or public figure, to sell a fake product or investment. Mark Cuban says ads using his likeness promote a $294 diabetes cure he has never sold. Real endorsements are rare for products like these, so treat the pattern as a warning sign.
How can you tell if a Facebook ad is a deepfake scam?
Look for a famous face promoting a miracle cure or guaranteed returns, pressure to act quickly, and a product you can't find anywhere else. Search the product name with "scam," and check the advertiser in Meta's Ad Library. If you're still unsure, don't pay, and report it at ic3.gov.
Why does Meta keep running scam ads?
According to Reuters' reporting on internal documents, Meta's automated systems ban an advertiser only when they're at least 95% sure it's a scam. Suspected advertisers below that line can keep running and pay more per ad. Meta disputes the way Reuters described this and says its scam ad removals and complaint numbers have improved.
Did Meta respond to Mark Cuban's claims?
Meta hasn't publicly confirmed Cuban's account, and Moneywise reported it didn't immediately get a response to his comments. Meta's public position in the Reuters reporting was that the internal documents gave a selective and distorted picture of its fraud work.
Protect Your Feed and the People in It
Facebook deepfake scam ads work because they exploit trust in faces and voices, and because enforcement thresholds leave room for profitable borderline advertisers. You can't fix Meta's system, but you can change what happens in your own feed. This week, pass this article to whoever in your family uses Facebook most, and ask them to search one suspicious product name with the word "scam" next to it. Then tell each other what came up.