Brazil: police say a group used deepfake videos of TV host Marcos Mion in fake restaurant promotions and victims paid for vouchers that did not exist
The Civil Police of Brazil's Federal District (PCDF) say a criminal group manipulated the image and voice of TV presenter Marcos Mion with artificial intelligence in fraudulent videos promoting fake discounts at the Outback restaurant chain. According to the police, people who clicked the advertisements were sent to cloned websites and paid for discount vouchers that were never delivered. In January 2025 the police's first phase of Operation VoiceOver arrested three suspects; outlets relaying the police reported dozens of victims and 27 complaints in four days. A second phase on 9 May 2025 arrested a man in Santa Helena de Goiás who, police say, built and maintained the cloned sites and confessed his role. Every account relays the police; no outlet examined the videos, no tool is named and no individual victim or total loss is described.
- AI system
- Unidentified video tool
- Occurred
- Event date unknown
- Reported
- 16 January 2025
- Event location
- Brazil
- What the AI did
- Depicted or impersonated the person
- Reported harm
- Financial Loss
- Whose AI use
- Someone else’s AI use
- Setting
- Finance · Everyday life
- Evidence
- AI involvement reported · Causal attribution alleged · 5 sources, 1 underlying account
- 4 claims: 4 reported. 4 open questions
- People reported harmed
- Not reliably quantifiable from the sources
AI system as recorded: Videos in which the image and voice of TV presenter Marcos Mion were manipulated with artificial intelligence ('deepfake'), according to the Federal District Civil Police, used in fake promotions for a restaurant chain; tool not identified
What Happened
According to the Civil Police of the Federal District (PCDF), relayed by Brazilian outlets in January and May 2025, a criminal group used videos manipulated with deepfake technology of the image and voice of TV presenter Marcos Mion to promote false promotions of a restaurant chain (Outback). A delegate of the police cybercrime unit (DRCC) said the group used artificial intelligence to emulate the presenter's voice and won the victims' trust through advertising campaigns. ICL Notícias reports that the group modified the image and audio of real videos of artists, that people who clicked the advertisement were sent to fake sites to pay for a 'promotional voucher', and that the suspects used cryptocurrency to launder the money. The police statement quoted by CNN Brasil says the victims were directed to the fake sites, where they paid for discount vouchers that did not exist. In the first phase of Operation VoiceOver in January 2025, three suspects were arrested in the Federal District; Seu Crédito Digital and Portal Tela, relaying the police, describe dozens of victims with losses still being calculated, and ICL Notícias reports that, according to the police, there were victims across the country and 27 people reported the scam in four days. In the second phase on 9 May 2025, a man was arrested at his home in Santa Helena de Goiás; the police say he developed the cloned sites and confessed his role. Outback said in a statement that it publicises promotions only on its official channels, and the presenter's office said any campaign involving him is announced only on his official channels and that legal measures were under way. No individual victim, amount lost or forensic finding about the videos is described in the sources read.
Reported harm
According to the Federal District Civil Police, people who clicked fake restaurant promotions using AI-manipulated videos of presenter Marcos Mion paid on cloned websites for discount vouchers that did not exist; outlets relaying the police describe dozens of victims, with losses not totalled.
How Harm Occurred
AI-manipulated videos of a well-known presenter gave fake restaurant promotions credibility; according to the police, people who clicked the advertisements were taken to cloned sites and paid for vouchers that did not exist. The generated likeness is the reported lure; the sites, payments and laundering were operated by the group.
Outcome
OngoingPhase 1 of Operation VoiceOver (January 2025, Federal District): three suspects arrested, vehicles and devices seized. Phase 2 (9 May 2025): a man arrested at his home in Santa Helena de Goiás under a temporary arrest warrant; searches, bank-account freezes; police say he confessed to creating and maintaining the fraudulent sites. Police say the suspects will answer for electronic fraud, criminal association and money laundering. No trial outcome is reported in the sources read.
What remains unknown
- The number of people who paid and the total amount lost; no individual victim is described.
- When the fake promotions ran and when victims paid; the sources date only the police operations (January and 9 May 2025).
- Which AI tool produced the videos and on which platforms the advertisements ran.
- Whether the three January suspects or the May suspect have been charged or tried.
What the evidence supports
AI involvement: reported. The Federal District Civil Police say the group used videos in which presenter Marcos Mion's image and voice were manipulated with deepfake technology, and a DRCC delegate says the group used artificial intelligence to emulate the presenter's voice and won the victims' trust through advertising campaigns; according to the police, people who clicked these advertisements were sent to cloned sites and paid for vouchers that did not exist. The AI-made likeness impersonated the presenter toward the people deceived (depicted_or_impersonated). No outlet examined the videos and no tool is named.
4 claims: 4 reported. What the statuses mean
Reported The Federal District Civil Police say a criminal group used videos in which presenter Marcos Mion's image and voice were manipulated with deepfake technology, using artificial intelligence, to promote false promotions of a restaurant chain.
Causal attribution. Police account (PCDF, DRCC delegate) relayed by several outlets; one reporting chain. No outlet examined the videos.
- cnnbrasil.com.br(opens in new tab) supports · Portuguese
'foi preso por integrar o grupo criminoso que utilizava vídeos manipulados com deepfake da imagem e voz do apresentador Marcos Mion para aplicar golpes'
- tribunadoplanalto.com.br(opens in new tab) supports · Portuguese
'O grupo criminoso utilizava vídeos manipulados com tecnologia deepfake da imagem e voz do apresentador Marcos Mion para promover falsas promoções de uma rede de restaurantes.'; 'os criminosos conseguiam simular a imagem e a voz de Marcos Mion, com uso de inteligência artificial'
- seucreditodigital.com.br(opens in new tab) supports · Portuguese
'Eles utilizavam a inteligência artificial para emular a voz do apresentador, atraindo a confiança das vítimas por meio de campanhas publicitárias'
Reported According to the police, victims who clicked the advertisements were directed to cloned websites, where they paid for discount vouchers that did not exist.
Causal attribution. Police statement; the same chain. Individual victims and amounts are not described.
- cnnbrasil.com.br(opens in new tab) supports · Portuguese
'As vítimas eram direcionadas aos sites falsos desenvolvidos pelo investigado, onde realizavam pagamentos por vouchers de desconto inexistentes', afirmou a polícia.
- iclnoticias.com.br(opens in new tab) supports · Portuguese
'Ao clicarem no anúncio, as vítimas eram direcionadas para sites falsos para concluírem o pagamento de um “voucher promocional”.'
- tribunadoplanalto.com.br(opens in new tab) supports · Portuguese
'As vítimas eram direcionadas aos sites falsos desenvolvidos pelo investigado, onde realizavam pagamentos por vouchers de desconto inexistentes, acreditando estar participando de promoções com a participação de Marcos Mion.'
Reported Outlets relaying the police in January 2025 reported dozens of victims, with losses still being calculated, and 27 complaints about the scam in four days.
Causal attribution. Police figures as relayed; the outlets differ on whether the victims are in the Federal District or across the country, and the 27 complaints are reports.
- iclnoticias.com.br(opens in new tab) supports · Portuguese
'De acordo com a polícia, há vítimas em todo o país. Em apenas quatro dias, 27 pessoas denunciaram o golpe.'
- portaltela.com(opens in new tab) supports · Portuguese
'As investigações revelaram que dezenas de vítimas foram lesadas, com prejuízos ainda em apuração.'
- seucreditodigital.com.br(opens in new tab) supports · Portuguese
'Dezenas de vítimas no Distrito Federal já foram identificadas, mas o número pode crescer à medida que as investigações avançam.'
Reported Three suspects were arrested in the first phase of Operation VoiceOver in January 2025, and on 9 May 2025 a man was arrested in Santa Helena de Goiás; police say he built and maintained the cloned sites and confessed.
Causal attribution. Police account; arrests and confession as stated by police, not a court finding.
- seucreditodigital.com.br(opens in new tab) supports · Portuguese
'A Polícia Civil do Distrito Federal (PCDF) desarticulou, nesta quarta-feira (15/1), uma quadrilha que utilizava tecnologia de deepfake para aplicar golpes digitais em nome do apresentador Marcos Mion.'
- iclnoticias.com.br(opens in new tab) supports · Portuguese
'Uma operação da Polícia Civil do Distrito Federal prendeu três suspeitos de usar Inteligência Artificial para aplicar golpes.'
- cnnbrasil.com.br(opens in new tab) supports · Portuguese
'a prisão aconteceu durante a segunda fase da Operação "VoiceOver", dando continuidade a ações ocorridas em janeiro'; 'O homem confessou a participação no esquema, durante o interrogatório, e detalhou seu papel em criar e manter os sites fraudulentos.'
Sources
5 sources inspected, from 1 underlying account. Sources that repeat one account do not corroborate each other.
- CNN Brasil, 9 May 2025: Integrante de grupo que fazia deepfake de Marcos Mion é preso em Goiás(opens in new tab)
s1 · cnnbrasil.com.br · News report · Portuguese · Inspected · 9 May 2025 · Shares an underlying account with another listed source · Primary
- ICL Notícias, 16 January 2025: Operação mira suspeitos de usar IA para aplicar golpes com imagem de famosos(opens in new tab)
s2 · iclnoticias.com.br · News report · Portuguese · Inspected · 16 January 2025 · Shares an underlying account with another listed source
- Portal Tela, 16 January 2025: Polícia desarticula quadrilha que usava deepfake de Marcos Mion para fraudes digitais(opens in new tab)
s3 · portaltela.com · News report · Portuguese · Inspected · 16 January 2025 · Shares an underlying account with another listed source
- Seu Crédito Digital, 16 January 2025: Golpe com Deepfake: Polícia do DF Prende Três Criminosos(opens in new tab)
s4 · seucreditodigital.com.br · News report · Portuguese · Inspected · 16 January 2025 · Shares an underlying account with another listed source
- Tribuna do Planalto, 9 May 2025: Grupo simulava voz de Marcos Mion para aplicar golpes cibernéticos(opens in new tab)
s5 · tribunadoplanalto.com.br · News report · Portuguese · Inspected · 9 May 2025 · Shares an underlying account with another listed source
How the sources were read, and where the events happened
Read in Portuguese and translated by the research agent (an AI) on 2026-10-11 (HTTP 200); no human translator or reviewer read the article. CNN Brasil, 9 May 2025 (second phase), quoting a Polícia Civil do Distrito Federal statement. Applies to s1.
Read in Portuguese and translated by the research agent (an AI) on 2026-10-11 (HTTP 200); no human translator or reviewer read the article. ICL Notícias, 16 January 2025 (first phase), relaying the police account. Applies to s2.
Read in Portuguese and translated by the research agent (an AI) on 2026-10-11 (HTTP 200); no human translator or reviewer read the article. Portal Tela, 16 January 2025 (first phase), relaying the police account. Applies to s3.
Read in Portuguese and translated by the research agent (an AI) on 2026-10-11 (HTTP 200); no human translator or reviewer read the article. Seu Crédito Digital, 16 January 2025 (first phase), quoting the DRCC delegate. Applies to s4.
Read in Portuguese and translated by the research agent (an AI) on 2026-10-11 (HTTP 200); no human translator or reviewer read the article. Tribuna do Planalto, 9 May 2025 (second phase), relaying the PCDF account. Applies to s5.
Event countries: Brazil. Affected-person countries: Brazil. Court countries: Unknown.
The investigation was run by the Civil Police of Brazil's Federal District, which arrested suspects in the Federal District and in Santa Helena de Goiás; ICL Notícias reports that, according to the police, there were victims across the country. No court proceeding is reported in the sources read.
Reviewed for publication 2026-10-11: One police reporting chain (Federal District Civil Police, relayed by CNN Brasil, ICL Notícias, Portal Tela, Seu Crédito Digital and Tribuna do Planalto) describes people who paid for nonexistent vouchers after fake promotions that, according to the police, used an AI-manipulated image and voice of a TV presenter. Published as a reported account with the count left unquantified.
People described
People in Brazil who paid for nonexistent discount vouchers after the fake promotions (not named or counted individually by any source)
People reported harmed in this case
Not reliably quantifiable from the sources
Police say victims paid for nonexistent vouchers but give no exact count: Seu Crédito Digital and Portal Tela relay 'dezenas de vítimas' (dozens, a vague plural that is not quantified here), and ICL Notícias reports 27 complaints in four days, which are reports rather than established losses. The presenter and the restaurant chain are not counted. Unquantified with zero placeholders.
Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.
Cite this case
Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.
APA
NOPE. (2026). Brazil: police say a group used deepfake videos of TV host Marcos Mion in fake restaurant promotions and victims paid for vouchers that did not exist. AI incidents. https://nope.net/incidents/2025-brazil-police-say-deepfake-marcos-mion-ads-sold-nonexistent-restaurant-vouchers
BibTeX
@misc{2025_brazil_police_say_deepfake_marcos_mion_ads_sold_nonexistent_restaurant_vouchers,
title = {Brazil: police say a group used deepfake videos of TV host Marcos Mion in fake restaurant promotions and victims paid for vouchers that did not exist},
author = {NOPE},
year = {2026},
howpublished = {AI incidents},
url = {https://nope.net/incidents/2025-brazil-police-say-deepfake-marcos-mion-ads-sold-nonexistent-restaurant-vouchers}
} Related cases
Brazil: scammers used AI-cloned voices and images of children with cancer in fake donation campaigns; 16 arrested in Operação Sophia
On 14 July 2026 the Rio Grande do Sul Civil Police launched Operação Sophia against a group that used AI voice cloning and deepfake to turn real fundraising campaigns of children with cancer into fake donation appeals: AI-generated child voices asked for money in videos boosted by pages such as 'Clube de Doadores' and 'Unidos pelo Amor', leading donors to lookalike sites that issued Pix codes to shell companies. Sixteen people were arrested across five states. In the campaign that started the investigation — using the image of Kelen Santos's daughter, who was in cancer treatment — police traced R$294,500 diverted, and the investigation found a financial-hub company that moved more than R$1.7 million. The family had to tell congratulating donors that their money had gone to a scam.
Paraíso das Águas, Mato Grosso do Sul: a farmer lost more than R$700,000, including a R$360,000 loan, after a video call in which scammers used AI to simulate the Viva Sorte prize-draw presenter
Three Mato Grosso do Sul outlets, all crediting BNC Notícias, reported on 10 April 2026 that a rural producer from Paraíso das Águas fell for the so-called Viva Sorte scam. In a video call, scammers used AI deepfake technology to simulate the image and voice of Renato Ambrósio, the presenter of the Viva Sorte prize draws, told the farmer a prize had been won and set out steps to release it: Pix transfers, sending the CPF tax number and sharing the phone screen while the banking app was open. The farmer followed the instructions and took out a R$360,000 loan; with interest, the total loss exceeded R$700,000. One report says the farmer gave this account to the police. The farmer is not identified, no date is given and no arrest is reported.
Bengaluru: a security guard says a WhatsApp video call in which an alleged deepfake 'Chief Minister Vijay' offered him financial aid led to fake 'processing' charges and a fake CBI fine, and he lost more than Rs 1.04 lakh; Byappanahalli police register an IT Act case (September 2026)
A security guard in Bengaluru, originally from Odisha, told the Times of India that on 9 September 2026 he answered a WhatsApp video call in which a face and voice presented as Tamil Nadu Chief Minister C. Joseph Vijay introduced himself in Hindi, asked his name, work and where he lived and pressed him to accept financial help. A 'manager' then promised Rs 11 lakh, said Rs 5 lakh had been allotted to him and asked for a Rs 5,000 exchange charge, then Rs 18,000 to unlock a supposedly locked PIN; a caller posing as a CBI officer demanded more than Rs 50,000 as a fine. The fraudsters sent a fake allotment letter and a purported CBI officer's identity proof. He paid more than Rs 1.04 lakh through a digital payment app before refusing a further Rs 50,000 demand, called the 1930 cybercrime helpline and went to Byappanahalli police, who registered a case under the Information Technology Act. The Times of India says the fraudster allegedly used a deepfake AI-generated video and calls it the first such case reported in the city. No arrest is reported, and no link to the Tamil Nadu CB-CID deepfake case or its Alwar arrest has been established.
India: AI deepfake videos of Tamil Nadu Chief Minister Vijay offering 'financial assistance' were used to defraud people in Madhya Pradesh and other states; a Rajasthan man was arrested
In 2026 AI-generated videos showing a simulated Tamil Nadu Chief Minister C. Joseph Vijay speaking in a fabricated Hindi voice circulated on Facebook and other platforms, urging viewers to contact a WhatsApp number to receive money for education, medical treatment and other needs. The Tamil Nadu Crime Branch-CID detected the videos during routine monitoring, registered a case on 1 September 2026 under the Bharatiya Nyaya Sanhita and the Information Technology Act, and traced the scheme to Rajasthan and Madhya Pradesh. On 12 September a 28-year-old man from Alwar district was arrested with Rajasthan Police, brought to Tamil Nadu on a transit warrant and remanded in judicial custody; a phone, SIM cards and a CPU were seized. Investigators traced a man in Madhya Pradesh who had lost money, and said people in several states were cheated. The CB-CID is also examining similar videos posted from Facebook accounts suspected to be run from Pakistan.
If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.