Short answer: A celebrity deepfake in an ad is an artificially generated video or recording in which a well-known person supposedly recommends an investment they never recommended. You can spot it by four groups of signals: errors in the image and voice, a promise of profit without risk, time pressure, and a suspicious site the ad leads to. The most important rule: a famous face is not proof, so check the offer in an independent source before you give any details.
You see an ad on Facebook: a well-known athlete, politician or CEO of a big company says “ordinary people” are earning thousands a week on a new platform. It looks credible, and the voice sounds familiar. That’s how a fake deepfake ad works, and in recent years it has affected hundreds of well-known people in Poland and around the world. In this guide you’ll find 15 documented examples, specific warning signs, a 60-second verification test and the story of Rafał Brzoska, a Polish entrepreneur who openly fights such ads.
What a deepfake in an ad is and why a famous face increases trust
A deepfake is material (video, image or audio) created or altered with artificial intelligence so that it shows a real person who never said or did anything of the kind. In investment scams, criminals clone a celebrity’s face and voice, sync the lip movements to an invented script and insert the recording into a template resembling a news program.
According to a May 2024 analysis by NASK, seconds of recording are enough to clone a voice, and more expressive speech can be produced from about a minute of material. So anyone who has ever spoken publicly can become a victim.
Four psychological mechanisms a deepfake relies on
- Borrowed authority. If someone famous and respected recommends it, the offer must be fine.
- The apparent credibility of the format. A news ticker, a portal’s logo and a TV studio are meant to suggest it’s genuine news coverage.
- Time pressure and exclusivity. “Access only for the first 100 people” shortens the time to think.
- Emotions instead of arguments. The promise of financial freedom appeals to dreams, not facts.
How the scam works step by step: from an ad to losing your savings
The deepfake itself is just the beginning. The real loss comes later and usually follows a similar pattern, described by CSIRT KNF and industry media, among others.
An ad with a celebrity deepfake
A paid social media ad promises quick, guaranteed profit and often looks like an article from a well-known news site.
A site imitating a professional platform
After clicking, you land on a page that asks for your name, phone number and email. Sometimes it’s a copy of the site of a well-known institution or event.
A call from an “advisor” or an invitation to a group
The scammer calls, persuades you to make a first deposit and often asks you to install a remote access program. Another variant is an “exclusive” WhatsApp or Telegram group with supposed tips. Vishing works the same way, as we write in the article Your bank is calling: 5 questions that expose vishing in 30 seconds.
Showcase “profits” and more deposits
On a fake dashboard, virtual profits grow to encourage you to deposit larger amounts.
Account takeover and disappearance
Once the victim is convinced, the scammer asks for login details to their account or another transfer. Then contact stops, and the money is hard to recover.
15 examples: deepfakes of famous people in fake ads
The table below collects documented cases from Polish and international sources (you’ll find links to them in the sources section). It shows that scammers use the image of politicians, athletes, journalists, doctors and even brands.
| # | Who or what | What the ad promised | How the scam was detected |
|---|---|---|---|
| 1 | Rafał Brzoska (CEO of InPost, a Polish parcel locker company) | Fake investments on Facebook (e.g., March 2024) | NASK analysis, later action by the Polish data protection authority (UODO) and a court (described below) |
| 2 | Robert Lewandowski | The “Immediate Matrix” platform, big profits in a few weeks (July 2024) | A mechanical voice without breathing, lip movements not matching speech, a fragment of a real interview, an account created on the day of publication |
| 3 | Donald Tusk (Polish Prime Minister) | An investment platform, PLN 30,000 a week (October 2025) | The lips moved after the sound ended, odd phrasing, the material came from a real speech about something completely different, a link to a fake site |
| 4 | Andrzej Duda (then Polish President) | Exceptionally profitable investments (May 2024) | NASK warning |
| 5 | Izabela Leszczyna (then Polish health minister) | Exceptionally profitable investments (May 2024) | NASK warning |
| 6 | Kamil Labudda (influencer) | Exceptionally profitable investments (May 2024) | NASK warning |
| 7 | Piotr Kraśko, Danuta Holecka, Wojciech Cejrowski (Polish TV journalists and presenters) | Calls to invest | NASK analysis from July 2024: journalists are the most often impersonated group (32% of cases) |
| 8 | Monika Olejnik, Paulina Smaszcz (Polish journalists) | Promoting a medicine | NASK analysis from July 2024 |
| 9 | Prof. Jerzy Szaflik (ophthalmologist) | Promoting a medicine he never recommended | He found out thanks to information from a patient |
| 10 | Omena Mensah (journalist) | Ads disguised as news, e.g., about her death or arrest | A complaint to UODO covering 263 ads |
| 11 | Żabka Polska (convenience store brand) | Quick earnings on an “investment platform” (October 2024) | CSIRT KNF warning |
| 12 | Martin Wolf (Financial Times journalist) | Fake investment offers (2025) | The avatar reached almost a million Meta users in the EU, according to media reports |
| 13 | Martin Lewis (financial expert, UK) | A supposed investment involving Elon Musk (July 2023) | Lewis himself warned that it was the first deepfake of him in a scam; Meta removed the ads |
| 14 | Taylor Swift | Free Le Creuset cookware “due to a packaging error” (January 2024) | Le Creuset denied any involvement; Meta removed the ad |
| 15 | Tom Hanks and MrBeast | A dental plan and an iPhone 15 Pro for $2 (October 2023) | Hanks himself warned fans on Instagram; the MrBeast video appeared on TikTok |
The common pattern in all the examples
Although the faces and languages differ, the pattern repeats: a famous person, a dramatic pretext (an error, an exceptional opportunity, “the elite’s secret”), a promise of quick profit or a free prize, and a link to a site where you have to enter your details. If you recognize this pattern, you’ll recognize the scam even when the deepfake is technically perfect.
How to spot a deepfake and a fake ad: 17 specific signals
A single signal isn’t conclusive, but several at once are very strong evidence of a scam. NASK stresses that the best defense is judging the content itself, not just how the video looks and sounds. That’s why we’ve divided the signals into four groups.
Group 1: signals in the image and sound
- Lip movements don’t match the words. In the ad featuring Donald Tusk, his lips kept moving after the sound had ended (Demagog).
- The voice is flat and mechanical. In Lewandowski’s case, Demagog pointed to the lack of audible breathing and unnatural sentence intonation.
- Blurring around the mouth, odd teeth, stiff facial expressions. NASK lists such artifacts as technical signs of a deepfake.
- Errors in numbers, amounts and names. A synthetic voice often mispronounces numbers, dates and proper names.
- Changes in tone or accent during speech and noise or “grain” laid over the image that masks imperfections.
Group 2: signals in what’s said
- Guaranteed profit with no risk. According to Poland’s Ministry of Digital Affairs, a promise of high profits with no risk of loss is a typical sign of fraud.
- Time pressure and limited places. Slogans like “only for the first 100 participants” are meant to switch off your thinking.
- A famous person talks about something they’ve never done. Tusk’s original speech didn’t contain a single word about investing, and Lewandowski didn’t mention any platform in the real interview.
- Odd or contradictory phrasing. The Tusk ad included, among other things, a claim about “access for 500 new users,” which clashed with the rest of the message.
Group 3: signals in the ad itself and the profile
- The ad pretends to be an article from a well-known news site. A fake news ticker, logo and studio are meant to build credibility, and the ministry explicitly describes this as a warning sign.
- The advertiser’s profile is new and empty. The account behind the Lewandowski ad was created the same day and had no other posts or reviews.
- The ad concerns something the person doesn’t promote on their official profiles. Real partnerships are usually visible on the celebrity’s official accounts too.
- Well-known brands as bait. Examples from Poland include Żabka, PKO Bank Polski, XTB and Orlen, whose names or logos were used in fake offers.
Group 4: signals on the landing page and after clicking
- The site’s address doesn’t match the advertised brand. In Tusk’s case, the link led to a site pretending to be an economic forum. We describe the rules for reading addresses in the article How to read a URL and domain before you click a link.
- A contact form instead of terms and company details. No tax ID, address or terms and conditions is a reason to leave the site.
- A call from an “advisor” pushing for a transfer. It’s a sign the contact is shifting from an ad into classic fraud.
- A request to install a remote access program or for your banking login details. A legitimate company shouldn’t require this.
| Signal | Why it’s a red flag | What to do |
|---|---|---|
| Profit without risk | Legitimate investments always involve risk | Close the ad and report it to the platform |
| Lip and voice errors | Typical artifacts of face and voice cloning | Watch the video again, muting and unmuting the sound |
| A fake news ticker | An imitation news site is meant to replace proof | Go to the real news site yourself and look for the story |
| A new advertiser account | Scammers create profiles just for a campaign | Check the creation date and the page’s history |
| An unrelated domain | The site doesn’t belong to the advertised brand | Check the address and the domain’s age |
| A call from an advisor | Pushing for a deposit is a typical stage of the scam | Hang up and don’t pay anything |
| Remote access | Gives the scammer control over your phone | Don’t install anything |
The 60-second test: how to check a suspicious deepfake ad
Before you click “Sign up,” follow the steps below. They’ll take you about a minute.
Stop and don’t click
The ad is designed to make you click on impulse. Wait a moment and ask yourself: does this famous person really recommend things like this?
Check the person’s official sources
Go to their verified profiles or website. If the partnership is real, you’ll find information about it. If not, you can also look at fact-checking services such as Demagog, which describe many such ads.
Use image search or a phrase from the ad
Search Google for a fragment of the ad’s slogan, or do a reverse image search of a frame from the video. That’s how Demagog found the original Tusk speech the deepfake was made from. We describe the methods in the article Fact-checking step by step.
Check the company, the site’s address and its history
Verify the domain, the company in the registers and the site’s archive. You’ll find a ready checklist in the article How to check whether an online store really exists. For investments, also check the official databases and warnings of your financial regulator (in Poland, the KNF).
Report it and don’t share it
Report the ad and profile to the platform and the fake site to CERT Polska (or your national equivalent). Poland’s Ministry of Digital Affairs also asks people not to share such offers, because you expose others to losses.
If you want to check whether a specific recording or photo was generated by AI, see the article Jak sprawdzić, czy zdjęcie, wideo lub tekst zostały wygenerowane przez AI.
Rafał Brzoska vs. fake deepfake ads: the story of one of the most high-profile fights
InPost CEO Rafał Brzoska is one of the faces most often used in fake ads in Poland. His image and that of his wife, journalist Omenaa Mensah, appeared in ads on Facebook and Instagram. Importantly, the couple didn’t stop at warnings: they took legal and public action that became a point of reference for the whole debate about platform responsibility. Below we describe the facts from documented sources. Some assessments, especially regarding the scale of the problem and the platform’s responsibility, come from Brzoska himself or his foundation, and Meta disagrees with them.
What the fake ads featuring Brzoska and Mensah looked like
According to media reports, the paid ads looked like news coverage. In some, a deepfake of Brzoska promoted fake investment platforms. Others featured made-up reports about Omenaa Mensah, including her arrest, assault or death. Brzoska wrote publicly that even his children ask him about such ads. In her complaint to UODO, Mensah identified 263 ads.
Legal steps: UODO, the court and a lawsuit
- August 5, 2024: The President of UODO, Mirosław Wróblewski, ordered Meta to stop displaying such content featuring Brzoska and Mensah in Poland for three months. Meta appealed the decisions to the Voivodeship Administrative Court in Warsaw, then withdrew the appeals, so the court discontinued the proceedings (information from March 2025).
- November 2024: Brzoska announced that a court had granted an injunction for the maximum period of one year. Under it, Meta was to face a penalty for every further deepfake ad featuring the couple.
- December 2024: Brzoska looked for other well-known people, including former presidents and prime ministers, to join legal action against Meta. He told the Financial Times about roughly 150 people whose image had been used in this way.
- Meta’s position: the company argued that the EU Digital Services Act (DSA) applies, and that the order would amount to general, proactive monitoring of content, which the DSA prohibits.
Brzoska’s ideas for fighting fake ads
In 2026, Brzoska moved from lawsuits to systemic proposals. In August he announced a grassroots, nonpartisan initiative against online fraud, modeled on his earlier deregulation campaign SprawdzaMY. The goal was to change the law so ordinary users could protect themselves and their families more effectively.
On September 21, 2026, as part of the “150 percent” initiative, he presented three specific proposals:
| Proposal | What it involves |
|---|---|
| 1. Verifying advertisers | Large platforms would check who buys an ad and who pays for it, according to the AML/KYC standards known from banking. For higher-risk entities, the ultimate beneficial owner would also be checked. An advertiser who can’t be reliably verified wouldn’t be able to launch a campaign. |
| 2. The 150% rule | If a platform runs a fraudulent ad despite verification and earns money from it, it would pay an administrative fine of 150% of the revenue from that ad and related campaigns. In other words, it gives back all the revenue and pays an extra 50%. |
| 3. A fund for victims | Part of the fines would go to a fund supporting people who lost money or data or suffered other harm as a result of fraud. |
According to media reports, a petition for tough sanctions gathered about 167,000 signatures. For now these are proposals: the Ministry of Digital Affairs said it was analyzing them, and the sources we use don’t mention a bill being submitted.
ScamWatch: a tool for detecting suspicious ads
The initiative also includes ScamWatch, announced at the end of August 2026. The tool, where users can report suspicious ads, analyzes ads in Meta’s ad library, links them to domains used by scammers and examines how quickly platforms respond. Experts from Rafał Brzoska’s Foundation work with the reports. According to a September 2026 report, more than 2 million ads were collected, of which 16.5% were flagged as suspicious and 1.6% were rated “very high risk.”
What this story teaches us
Three practical lessons come from the Brzoska case. First, even a person with big resources and a team of lawyers can’t quickly remove all the fake ads. Second, responsibility for ads is a dispute about law, not just about technology. Third, until the rules change, you remain the first line of defense.
What to do after seeing a fake deepfake ad and after a possible payment
As soon as you see a suspicious ad
- Don’t click the link and don’t give any details.
- Report the ad and the profile to the platform’s administrators (Meta, Google).
- In Poland, report the fake site to CERT Polska via the form incydent.cert.pl/domena or in the mObywatel app under “Safe online” (elsewhere, use your national cybercrime reporting service).
- Don’t share the ad, even to warn others. Instead, describe the problem in your own words and add a link to an explanation.
If you’ve already given details or paid money
- Call your bank immediately and ask them to block your cards and try to stop the transaction.
- If you installed a remote access program, uninstall it and change your passwords from another device.
- Report it to the police and keep the evidence: screenshots, the site’s address, phone numbers and transfer confirmations.
- Consider freezing your identity (in Poland, blocking your PESEL number; elsewhere, a credit freeze) if you gave personal details, to make it harder to take out loans in your name.
You’ll find more about what to do step by step after talking to a scammer in the article on vishing linked above.
How to help loved ones, especially older people
Show your parents or grandparents 2–3 examples from the table above, ideally with faces they know from TV. Say plainly: “no famous person calls you up and recommends guaranteed profits.” Agree on one family rule: before anyone puts money into an investment found online, they show the offer to someone they trust. For younger family members, it’s worth adding that even a video of their “favorite YouTuber” may be fake.
Frequently asked questions about deepfakes in ads
It’s a video, image or recording generated by artificial intelligence in which a real person supposedly promotes a product or investment, although they never did. Scammers use a cloned voice and face of a well-known person for this.
Look for a promise of profit without risk, time pressure, errors in lip movement and voice, a new or empty advertiser account and a link to an unknown site. Also check whether the person mentions such a partnership on their official profiles.
No. According to Poland’s Ministry of Digital Affairs, a promise of high profits without risk of loss is a typical sign of fraud. A famous face and the logo of a big institution don’t prove that the person or company endorses the offer.
Report the ad and profile to the platform’s administrators, e.g., Meta or Google. In Poland, report the fake site to CERT Polska via the form at incydent.cert.pl/domena or in the mObywatel app under “Safe online”; elsewhere, use your national cybercrime reporting service.
Call your bank immediately, report it to the police, keep the evidence, uninstall any remote access programs and change your passwords. If you gave personal details, consider freezing your identity or credit.
As part of the “150 percent” initiative, Brzoska proposes mandatory verification of advertisers according to AML/KYC standards, a fine for platforms equal to 150% of the revenue from a fraudulent ad, and a fund paying compensation to victims, financed from part of the fines. For now these are proposals, not laws in force.
That’s the subject of a legal dispute. Meta cites the EU Digital Services Act (DSA) and argues that an order for proactive monitoring would be unlawful. The other side, including Brzoska, demands greater platform responsibility. The matter is ongoing, so for now your own caution is what matters most.
Want to search and verify even more wisely?
Get the “How to Search?” guide
Practical methods for searching and verifying information, step by step, with dozens of examples.
See the guide → ☕Support us on Buy Me a Coffee
We create these materials in our spare time. Every bit of support helps us write more, and more often.
Support → ▶️Watch us on YouTube
The same topics in a shorter video format — with concrete examples on screen.
Subscribe →More from our blog
Sources
- NASK, “NASK warns: deepfake scams using images of famous people are on the rise” (May 2024): nask.pl
- CyberDefence24, “Deepfake. Scammers impersonate journalists, politicians, athletes” (July 2024, NASK analysis): cyberdefence24.pl
- Demagog, “Is Tusk advertising an investment platform? It’s a deepfake” (October 3, 2025): demagog.org.pl
- Demagog, “Lewandowski encouraging people to invest? Another AI scam” (July 11, 2024): demagog.org.pl
- Bankier.pl, “Fake photos and videos. How AI helps scammers impersonate experts and celebrities” (September 23, 2026): bankier.pl
- XYZ.pl, “Fake investments, real faces. This is how scammers steal Poles’ trust” (October 14, 2025): xyz.pl
- Antyweb, on fake ads using the Żabka brand, CSIRT KNF notice (October 18, 2024): antyweb.pl
- Poland’s Ministry of Digital Affairs, “An ad with a famous person? Watch out, it could be a scam” (April 17, 2025): gov.pl
- TechCrunch, on the Martin Lewis deepfake (July 7, 2023): techcrunch.com
- NBC News, „That Taylor Swift AI-generated Le Creuset ad is not real”: nbcnews.com
- Synthedia, “MrBeast and Tom Hanks Followers the Latest Victims of Deepfake Scams”: synthedia.substack.com
- Fortune, on the MrBeast and Tom Hanks deepfakes (October 4, 2023): fortune.com
- Trend Micro, on the fake Taylor Swift Le Creuset ad (January 12, 2024): trendlife.com
- Prawo.pl, “Rafał Brzoska and Omenaa Mensah – UODO acts on the ads” (March 24, 2025): prawo.pl
- Bezprawnik, on Brzoska’s dispute with Meta and the court injunction (November 21, 2024): bezprawnik.pl
- Wirtualnemedia.pl, on the search for celebrities for a joint lawsuit (December 5, 2024): wirtualnemedia.pl
- Proto.pl, on Brzoska’s new initiative against online fraud (August 20, 2026): proto.pl
- XYZ.pl, on ScamWatch and the scale of fake ads: xyz.pl
- Portal Medialny, on the proposed legal changes and platform responsibility (September 23, 2026): portalmedialny.pl
- RMF24, “A fake ad should cost dearly. Brzoska proposes new sanctions” (September 2026): rmf24.pl
- Strefa Biznesu, on the end of anonymous ads and a fund for victims (September 21, 2026): strefabiznesu.pl
Descriptions of the positions of Brzoska, Meta and ScamWatch come from statements by the parties to the dispute as reported by the media; they are not final court findings. This article is for information only and is not legal or investment advice.