{"meta":{"exportedAt":"2026-11-07T02:12:23.287Z","formatVersion":2,"selection":{"q":"brazil","system":"","harm":"","context":"","country":"","role":"","relation":"","evidence":"","year":"2025","response":"","severity":"","verification":"","view":"incidents","sort":"added"},"totalIncidents":4,"coverage":{"cases":4,"countries":1,"languages":2,"unknownLocation":2,"locationPending":0,"unknownLanguage":0,"unknownDate":0,"lawsuits":1,"regulatory":1,"minors":0,"coreRelations":3,"contextualRelations":1,"mixedRelations":0,"unknownRelations":0,"relationPending":0,"relationUnknown":0},"countingNote":"Distinct public cases in this selection. People counts apply within individual cases only; cross-case person overlap has not been resolved. No population incidence estimate.","affectedCountNote":"Interpret person counts with affectedCountStatus and the reported effects. Unquantified zeros are placeholders, not a measured zero.","source":"AI incidents","publisher":"NOPE","url":"https://nope.net/incidents","license":"CC BY 4.0"},"incidents":[{"id":"2025-brazil-police-say-fake-online-donation-campaigns-used-ai-deepfake-videos-of-a-singer-and-two-priests","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'Segundo a Polícia Civil, o grupo criava e divulgava páginas falsas de arrecadação de doações' ... 'Para ampliar o alcance das campanhas e atrair mais pessoas, os investigados recorriam a ferramentas de inteligência artificial, incluindo as chamadas' ... 'para criar vídeos falsos com imagens de personalidades públicas, como Gusttavo Lima e os padres Marcelo Rossi e Fábio de Melo.' ... 'As publicações circulavam nas redes sociais e direcionavam os usuários a páginas que simulavam campanhas beneficentes legítimas. Os valores arrecadados, porém, eram transferidos para contas bancárias controladas pelo grupo.'","relation":"supports","source_id":"s1"},{"locator":"'Para dar credibilidade às campanhas, os suspeitos também recorriam a ferramentas de inteligência artificial, incluindo a tecnologia deepfake, para produzir vídeos falsos com imagens de personalidades como o cantor Gusttavo Lima e os padres Marcelo Rossi e Fábio de Melo.'","relation":"supports","source_id":"s2"}],"assertion":"According to the police, the group created and spread fake donation pages and, to attract donors, used artificial intelligence tools including deepfakes to make fake videos with the images of the singer Gusttavo Lima and the priests Marcelo Rossi and Fábio de Melo; the posts circulated on social media and led users to pages imitating legitimate charity appeals, and the money went to accounts the group controlled.","causal_attribution":"Police account relayed by the outlets; all sources relay one police release, and the technical basis for the AI attribution is not disclosed."},{"id":"c2","status":"reported","evidence":[{"locator":"'Os investigados usavam inteligência artificial para criar vídeos de falsos idosos, crianças e pessoas com doenças graves para sensibilizar e induzir as vítimas a fazerem doações.'","relation":"supports","source_id":"s3"}],"assertion":"g1 reports that the suspects also used artificial intelligence to make videos of fake elderly people, children and seriously ill people to move victims to donate.","causal_attribution":"g1's account of the police investigation; the other reports describe these as images used on the pages without saying they were AI-made."},{"id":"c3","status":"reported","evidence":[{"locator":"'Entre 13 de outubro e 26 de novembro de 2025, a companhia recebeu 215.575 transações, que somaram R$ 9.839.087,73. A maioria dos pagamentos envolvia quantias relativamente baixas, especialmente R$ 30 e R$ 50.'","relation":"supports","source_id":"s2"},{"locator":"'O padrão aplicado pelo grupo sugere um número expressivo de vítimas em diferentes regiões do país, ainda que a soma de transações não corresponda necessariamente ao número de pessoas afetadas.'","relation":"supports","source_id":"s1"},{"locator":"'A PCDF ressalva que não dá para ligar esse total a um número exato de vítimas.'","relation":"supports","source_id":"s4"}],"assertion":"Police say one company linked to the group received 215,575 transactions totalling R$9,839,087.73 between 13 October and 26 November 2025, mostly of R$30 and R$50, and that the number of transactions does not necessarily match the number of victims.","causal_attribution":"Police financial findings relayed by the outlets."},{"id":"c4","status":"reported","evidence":[{"locator":"'A ação cumpre seis mandados de prisão preventiva, além de mandados de busca e apreensão no Distrito Federal e nos estados de Goiás, Minas Gerais, Santa Catarina e São Paulo. A Justiça também determinou o bloqueio de ativos financeiros e patrimoniais. Até o momento, cinco investigados foram presos.'","relation":"supports","source_id":"s1"},{"locator":"'Os agentes prenderam seis pessoas: duas em Planaltina, no Distrito Federal; uma em Planaltina de Goiás, uma em Uberlândia, Minas Gerais; e uma em Joinville, em Santa Catarina.'","relation":"supports","source_id":"s3"}],"assertion":"On 28 October 2026 the Federal District Civil Police launched an operation with six preventive arrest warrants, search warrants in the Federal District and four states, and asset freezes; five suspects had been arrested by the time of the CNN Brasil report, while g1 reported six arrests.","causal_attribution":"Procedural facts reported from the police; the arrest count differs between reports."},{"id":"c5","status":"reported","evidence":[{"locator":"'A análise de contas eletrônicas também revelou mais de 230 GB de arquivos, incluindo vídeos, imagens, roteiros, códigos de programação, bancos de dados e materiais destinados à produção de novas campanhas fraudulentas.'","relation":"supports","source_id":"s1"}],"assertion":"Investigators found more than 230 GB of files in the group's accounts, including videos, images, scripts, programming code, databases and material for new fraudulent campaigns.","causal_attribution":"Police finding relayed by CNN Brasil."},{"id":"c6","status":"reported","evidence":[{"locator":"'Os investigados poderão responder por estelionato eletrônico, organização criminosa e lavagem de dinheiro, entre outros crimes que venham a ser identificados no decorrer da apuração.'","relation":"supports","source_id":"s1"},{"locator":"'A PCDF não divulgou os nomes dos presos, e a defesa não foi localizada. A polícia descreve o cantor e os padres como imagens usadas nos vídeos falsos, não como participantes.'","relation":"supports","source_id":"s4"}],"assertion":"Police say the suspects may answer for electronic fraud, criminal organisation and money laundering; the police did not release the names of those arrested and describe the singer and the priests as images used in the fake videos, not as participants.","causal_attribution":"Police statements relayed by the outlets."}],"effects":[{"label":"people moved by fake charity appeals using AI-made videos of public figures sent donations to the group","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.cnnbrasil.com.br/nacional/centro-oeste/df/policia-prende-grupo-que-usava-deepfake-de-gusttavo-lima-em-golpe-de-doacao/","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-29 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. CNN Brasil report, read in full; it relays the PCDF operation release.","independence_group":"pcdf-operacao-caritas-release"},{"id":"s2","url":"https://jornaldebrasilia.com.br/brasilia/pcdf-prende-suspeitos-de-arrecadar-quase-r-10-milhoes-com-falsas-campanhas-de-doacao/","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-29 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Jornal de Brasília report, read in full; it relays the same PCDF release (photo credited PCDF).","independence_group":"pcdf-operacao-caritas-release"},{"id":"s3","url":"https://g1.globo.com/df/distrito-federal/noticia/2026/10/28/grupo-suspeito-ia-falsas-campanhas-beneficentes.ghtml","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-29 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. g1 DF report, read in full; it relays the police account and a police video; it gives six arrests where the other reports give five.","independence_group":"pcdf-operacao-caritas-release"},{"id":"s4","url":"https://soubrasilia.com/golpe-vaquinha-falsa-pix-deepfake-pcdf-operacao-caritas-como-evitar/","kind":"news_report","access":"read","language":"pt","translation_note":"Read in Portuguese by the research agent (an AI) on 2026-10-29 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. SouBrasília explainer, read in full; it relays the PCDF release and states the police caveat on victim numbers.","independence_group":"pcdf-operacao-caritas-release"}],"version":1,"ai_roles":["others_use"],"contexts":["finance","everyday_life"],"unknowns":["Which AI tools were used and how the police established that the videos were AI-made.","How many people donated, and how many did so because of the AI-made videos of public figures rather than the other material on the pages.","Whether the singer or the priests depicted have responded to the use of their images.","The full period of the campaigns beyond the 13 October to 26 November 2025 window traced for one company, and what charges will follow."],"geography":{"basis":"The police operation ran in the Federal District, Goiás, Minas Gerais, Santa Catarina and São Paulo, and the police say the victims were in different regions of the country; the warrants were issued by a Brazilian court.","court_countries":["BR"],"event_countries":["BR"],"affected_person_countries":["BR"]},"publication":{"basis":"Published as a core case (depicted_or_impersonated): the investigating Federal District Civil Police say the group used AI tools, including deepfakes, to make fake videos with the images of a singer and two priests to draw donors to fake charity pages, and that the group took in about R$9.8 million through one company in late 2025. The police's technical basis is not disclosed. All four sources relay one police release, so every claim is reported. No private person is named.","reviewed_on":"2026-10-29"},"ai_involvement":{"basis":"According to the Federal District Civil Police, whose release CNN Brasil, Jornal de Brasília, g1 DF and SouBrasília relay, the investigation found that the group used artificial intelligence tools, including deepfakes, to make fake videos with the images of a singer and two priests, and that these videos were used to attract donors to fake charity pages. This is an attribution of this scheme by the investigating police, which also report seizing more than 230 GB of the group's videos, images, scripts and code; how the police established that the videos were AI-made is not disclosed. People deceived by the generated videos of public figures were impersonated toward them (depicted_or_impersonated).","status":"reported"},"person_relations":["depicted_or_impersonated"]},"name":"Brasília police say a group used AI deepfake videos of singer Gusttavo Lima and two priests in fake donation drives that took in nearly R$10 million","summary":"The Federal District Civil Police (PCDF) in Brazil say a criminal group ran fake online donation campaigns that used images of elderly people, children and seriously ill people, and used artificial intelligence tools, including deepfakes, to make fake videos with the images of the singer Gusttavo Lima and the Catholic priests Marcelo Rossi and Fábio de Melo. The posts circulated on social media and sent people to pages imitating legitimate charity appeals; the money went to bank accounts the group controlled. Police say one company linked to the group received 215,575 transactions totalling about R$9.8 million between 13 October and 26 November 2025, mostly of R$30 and R$50, and that the number of transactions does not necessarily match the number of victims. On 28 October 2026 the PCDF launched Operação Cáritas with six preventive arrest warrants and search warrants in the Federal District and four states; five suspects had been arrested by the time of most reports (g1 reported six). How the police established that the videos were AI-made is not disclosed, and no response from the people depicted is reported.","incidentDate":"2025-10-13","incidentEndDate":"2025-11-26","incidentKind":"bounded_series","incidentDatePrecision":"range","exposurePattern":"single_interaction","reportedDate":"2026-10-28","aiSystem":"Artificial intelligence tools, not named, including deepfake technology, that the Federal District Civil Police say the group used to make fake videos with the images of a singer and two priests for fake donation campaigns","aiProduct":"Unidentified video tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["financial_loss"],"harmOutcomeSummary":"Police say people who saw fake charity appeals promoted with AI-made videos of a singer and two priests sent donations to accounts the group controlled; one linked company received about R$9.8 million in 215,575 transactions in October and November 2025.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":0,"affectedCountStatus":"unquantified","affectedCountEvidence":"Unquantified. The police describe victims who were induced to transfer money and say the transaction pattern suggests many victims, but they state that the 215,575 transactions do not necessarily correspond to the number of people affected, and no count of people is given. The 66 police reports in the Federal District concern companies receiving money from various electronic frauds and are not counted. The singer and priests depicted are not counted because no response or consequence is reported for them.","victimAgeRange":"unknown","jurisdiction":"BR","platformType":"other","outcomeType":"investigation_opened","outcomeStatus":"ongoing","primarySourceUrl":"https://www.cnnbrasil.com.br/nacional/centro-oeste/df/policia-prende-grupo-que-usava-deepfake-de-gusttavo-lima-em-golpe-de-doacao/","primarySourceLabel":"CNN Brasil, 28 Oct 2026: police arrest group that used Gusttavo Lima deepfake in donation scam (Portuguese)","firstPublishedAt":"2026-10-29T03:08:23.542533+00:00","updatedAt":"2026-10-29T03:08:23.542533+00:00","scopeVersion":"facts-v3","tags":["deepfake","donation-scam","fraud","police-investigation","public-figures","brazil","depicted-or-impersonated"]},{"id":"2025-brazil-telegram-group-ai-fake-intimate-images-court-orders-platform-damages","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'O Telegram foi condenado a indenizar em R$ 10 mil uma mulher que teve divulgadas imagens íntimas falsas, criadas com uso de inteligência artificial (IA).'; 'da 32ª Vara Cível de Goiânia'; 'A decisão foi assinada na última terça-feira (06)'","relation":"supports","source_id":"s1"},{"locator":"'A Justiça condenou a plataforma de mensagens Telegram a pagar R$ 10 mil por danos morais a uma mulher que teve divulgadas imagens íntimas falsas, criadas com uso de inteligência artificial (IA).'","relation":"supports","source_id":"s2"}],"assertion":"In a judgment signed on Tuesday 6 October 2026, the 32ª Vara Cível de Goiânia ordered Telegram to pay R$10,000 in moral damages to a woman whose fake intimate images, created with artificial intelligence, were disseminated on the platform.","causal_attribution":"The judgment as reported; the judgment itself was not inspected."},{"id":"c2","status":"reported","evidence":[{"locator":"'Usuários do Telegram usavam ferramentas de IA para modificar fotos publicadas pela mulher e pelas irmãs dela nas redes sociais. Eles criavam imagens de nudez e situações sexuais que não existiam, sem o consentimento delas.'; 'compartilhavam essas montagens em um grupo com mais de 120 participantes, tecendo comentários ofensivos e de conteúdo sexual às vítimas.'","relation":"supports","source_id":"s1"},{"locator":"'produzir montagens de cunho sexual com o uso de inteligência artificial'","relation":"supports","source_id":"s2"}],"assertion":"According to the judgment as reported, Telegram users used AI tools to alter photos the woman and her sisters had posted on social media into nude and sexual images without their consent, and shared them in a group of more than 120 participants with offensive sexual comments.","causal_attribution":"The court's account of how the images were made and shared, as reported; no tool is named."},{"id":"c3","status":"reported","evidence":[{"locator":"'por meio de formulário de suporte, mensagens eletrônicas e atendimento fornecido dentro do aplicativo.'; 'Em uma notificação enviada em 29 de julho de 2025, por exemplo, informou que a imagem dela estava sendo usada sem autorização.'; 'Contudo, não recebeu resposta nem conseguiu interromper a divulgação.'; 'Naquela época, tinha 98 participantes, mas chegou a ter pelo menos 123 um mês depois, no dia 27 de agosto.'","relation":"supports","source_id":"s1"}],"assertion":"She reported the group to Telegram through its support form, emails and in-app support, including a notification on 29 July 2025, and received no reply; the group had 98 participants then and at least 123 on 27 August.","causal_attribution":"Facts recorded in the judgment as reported by one outlet."},{"id":"c4","status":"reported","evidence":[{"locator":"'Ele explicou que o Telegram foi responsabilizado porque não removeu as postagens mesmo depois de notificado.'; 'A empresa também deve usar ferramentas de identificação de arquivos e reconhecimento de imagens para impedir a republicação de cópias idênticas ou semelhantes ao conteúdo ilegal.'","relation":"supports","source_id":"s1"},{"locator":"'que também determinou a retirada do material da plataforma, a adoção de mecanismos para impedir novas publicações e o fornecimento de dados que permitam identificar os usuários envolvidos'","relation":"supports","source_id":"s2"}],"assertion":"The court held Telegram liable because it did not remove the posts after being notified, and ordered it to remove the material, adopt mechanisms against new publication, and provide data identifying the users involved.","causal_attribution":"The court's reasoning and orders as reported."},{"id":"c5","status":"reported","evidence":[{"locator":"'Não se trata de mero uso indevido de imagem, mas da fabricação de uma intimidade sexual inexistente, que se apropria da identidade visual da vítima para expô-la, de forma degradante, a um número indeterminado de pessoas'","relation":"supports","source_id":"s1"}],"assertion":"The judge wrote that the case was not mere misuse of an image but the fabrication of a non-existent sexual intimacy that appropriates the victim's visual identity to expose her degradingly to an indeterminate number of people.","causal_attribution":"The judge's words as quoted by one outlet."},{"id":"c6","status":"reported","evidence":[{"locator":"'O Telegram foi citado, mas não contestou a ação.'","relation":"supports","source_id":"s1"},{"locator":"'O POPULAR entrou em contato com a defesa do Telegram sobre a decisão judicial, mas, até a última atualização desta reportagem, não obteve retorno.'","relation":"supports","source_id":"s2"}],"assertion":"Telegram did not contest the action and had not answered O Popular's request for comment.","causal_attribution":"Absence of a response from the platform; no contrary account is reported."}],"effects":[{"label":"fake nude and sexual images of her shared in a Telegram group of more than 120 participants","claim_id":"c2","direction":"negative"}],"sources":[{"id":"s1","url":"https://portal6.com.br/2026/10/13/telegram-e-condenado-em-goiania-apos-usuarios-compartilharem-fotos-intimas-de-mulheres-feitas-com-ia/","kind":"news_report","access":"read","language":"pt","translation_note":"Read on 2026-10-19 through an Internet Archive capture (the live page returned HTTP 403; bodies/lat-portal6-telegram-wb.html/.txt). Portal 6's report of the judgment, citing the legal news site Rota Jurídica, which was not read. Read in the original Portuguese; the English renderings in claims and notes are translations by the research agent (an AI), not by a human translator.","independence_group":"goiania-32vc-judgment-2026-10-06"},{"id":"s2","url":"https://opopular.com.br/cidades/telegram-e-condenado-em-goiania-por-compartilhamento-de-imagens-intimas-de-mulheres-criadas-com-ia-1.3468089","kind":"news_report","access":"read","language":"pt","translation_note":"Read on 2026-10-19 (live, HTTP 200; bodies/lat-opopular-telegram.html/.txt). O Popular's report of the judgment; the page is paywalled (isAccessibleForFree false) and only the standfirst and the first two paragraphs were readable. Read in the original Portuguese; the English renderings in claims and notes are translations by the research agent (an AI), not by a human translator.","independence_group":"goiania-32vc-judgment-2026-10-06"}],"version":1,"ai_roles":["others_use"],"contexts":["privacy","justice"],"unknowns":["Which AI tools were used and who made the images.","When the images were first made and shared; the reports give a notification on 29 July 2025 and the group's size then and on 27 August.","How many sisters were depicted.","Whether the users responsible have been identified or prosecuted, and whether Telegram has complied or appealed.","Whether the judgment itself finds that AI tools were used or records the plaintiff's uncontested account; the judgment text was not inspected."],"geography":{"basis":"The judgment was given by the 32ª Vara Cível de Goiânia in Goiás, Brazil, which supports court_countries=BR. The reports do not say where the woman lives or where the group's members were; the court's seat is not used to locate the event or the woman.","court_countries":["BR"],"event_countries":[],"affected_person_countries":[]},"publication":{"basis":"Two Goiás outlets report a civil judgment finding that Telegram users altered a woman's photos into fake sexual images with AI tools and shared them in a group, and that Telegram, after ignoring her reports, must pay damages and remove and block the images. The facts rest on the judgment as reported; the judgment itself was not inspected.","reviewed_on":"2026-10-19"},"ai_involvement":{"basis":"The judgment, as reported by Portal 6 and O Popular, says Telegram users used AI tools to alter photos the woman and her sisters had posted into nude and sexual images that did not exist, and the court treated the images as fake intimate images created with AI. The generated images depicting her are the harm. O Popular's standfirst says the court understood that the group used AI to create the fake sexual montages. The tools are not named, the judge's quoted words describe fabricated images without naming AI, Telegram did not contest the action, and the reports do not say how the court established that AI was used.","status":"reported"},"person_relations":["depicted_or_impersonated"]},"name":"Brazilian court orders Telegram to pay a woman damages after it ignored her reports of a group sharing AI-made fake intimate images of her","summary":"A civil court in Goiânia, Brazil (32ª Vara Cível) ordered Telegram in a judgment signed on 6 October 2026 to pay R$10,000 in moral damages to a woman whose fake intimate images, created with artificial intelligence, were shared on the platform. According to the judgment as reported by Portal 6 (citing the legal news site Rota Jurídica) and O Popular, Telegram users used AI tools to alter photos that she and her sisters had posted on social media into nude and sexual images, and shared them without consent in a group of more than 120 participants with offensive sexual comments. She reported the group to Telegram through its support channels, including a notification on 29 July 2025, and received no reply; the group had 98 participants then and at least 123 a month later. The court held Telegram liable because it did not remove the posts after notification, ordered removal of the material, measures to block identical or similar copies, and disclosure of data identifying the group's creators and users. Telegram did not contest the case and did not answer O Popular.","incidentDate":"2025-01-01","incidentKind":"bounded_series","incidentDatePrecision":"year","exposurePattern":"unknown","reportedDate":"2026-10-13","aiSystem":"AI tools used by members of a Telegram group to alter the woman's photos into fake nude and sexual images, according to the judgment as reported; no tool is named","aiProduct":"Unidentified image tool","severity":"medium","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["exploitation_or_abuse","reputational_harm"],"harmOutcomeSummary":"According to the judgment as reported, fake nude and sexual images of the woman made with AI tools were shared without consent in a Telegram group of more than 120 participants with offensive sexual comments, and kept circulating after she reported them. The court found a violation of her personality rights and awarded R$10,000 in moral damages.","frameworkFacets":[],"pathwayDirection":"outward","causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"partial","affectedCountEvidence":"One counted person, the woman who brought the case. Her sisters' photos were also altered and shared, but the reports do not say how many sisters; they are not counted. Group participants are an audience and are not counted.","victimAgeRange":"adult","jurisdiction":"BR","platformType":"other","outcomeType":"lawsuit_ongoing","outcomeStatus":"ongoing","primarySourceUrl":"https://portal6.com.br/2026/10/13/telegram-e-condenado-em-goiania-apos-usuarios-compartilharem-fotos-intimas-de-mulheres-feitas-com-ia/","primarySourceLabel":"Portal 6, 13 Oct 2026: Telegram é condenado em Goiânia após usuários compartilharem fotos íntimas de mulheres feitas com IA","firstPublishedAt":"2026-10-19T03:19:28.722651+00:00","updatedAt":"2026-10-19T03:19:28.722651+00:00","scopeVersion":"facts-v3","tags":["ai-generated-images","non-consensual-intimate-imagery","telegram","platform-liability","civil-judgment","brazil"]},{"id":"2025-br-meu-inss-rural-worker-retirement-claim-rejected","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"instantly turned down because the system identified","relation":"supports","source_id":"s1"},{"locator":"teve seu pedido negado de forma automática devido a um erro de identificação, sendo registrada como homem no sistema","relation":"supports","source_id":"s2"}],"assertion":"In a retirement claim filed through the Meu INSS app, the request was rejected because the system identified the claimant as a man.","causal_attribution":"Rest of World reports the rejection as caused by the system's identification of the claimant as a man. The source is the claimant's account as reported by the journalist. The Portuguese rewrite depends on the same article. No decision record was seen."},{"id":"c2","status":"reported","evidence":[{"locator":"in February was approved in March","relation":"supports","source_id":"s1"},{"locator":"went straight to INSS directors, who identified and corrected the mistake in the app.","relation":"supports","source_id":"s1"}],"assertion":"The rejected retirement claim was filed in February and approved in March after INSS directors corrected the mistake in the app.","causal_attribution":"The article ties the approval to the claimant's contact at an agricultural workers' confederation, and states that the case went straight to INSS directors. It does not say the contact routed it."},{"id":"c3","status":"reported","evidence":[{"locator":"had led to numerous rejections, with few options for recourse","relation":"supports","source_id":"s1"},{"locator":"proving everything, and [the benefit] still gets denied. It’s a humiliation,","relation":"supports","source_id":"s1"}],"assertion":"Rest of World reports that minor errors in the claimant's claims filed through the app led to numerous rejections, and quotes the claimant saying that despite having all the documents proving a health condition the benefit is still denied.","causal_attribution":"The article attributes the earlier rejections to minor errors in claims filed through the app. Dates of these rejections and whether an AI component made them are not stated."},{"id":"c4","status":"reported","evidence":[{"locator":"Each automated decision is based on specified legal criteria, ensuring that the standards set by the social security legislation are respected,","relation":"supports","source_id":"s1"},{"locator":"Some policyholders have been misusing the program, filing multiple requests in the hopes of obtaining different results, the spokesperson said.","relation":"supports","source_id":"s1"}],"assertion":"An INSS spokesperson told Rest of World that each automated decision follows specified legal criteria, and that some policyholders file multiple requests hoping for different results.","causal_attribution":"This is the institution's stated position and does not address the claimant's specific rejection."}],"effects":[{"label":"Retirement claim rejected, approved after INSS directors corrected the error","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://restofworld.org/2025/brazil-ai-social-security-app-rejected/","kind":"news_report","access":"read","language":"en","translation_note":"Full body read from a Wayback capture (20260128230439id_) after the live URL returned HTTP 429. Original English.","independence_group":"restofworld-daros"},{"id":"s2","url":"https://olhardigital.com.br/2025/04/25/pro/sistema-de-ia-do-inss-causa-recusas-injustas-de-beneficios/","kind":"news_report","access":"read","language":"pt","translation_note":"Portuguese rewrite of the Rest of World article (it credits Rest of World). Read in the original Portuguese by the reviewing agent without a human translator. Adds no independent reporting.","independence_group":"restofworld-daros"}],"version":1,"ai_roles":["institutional_use"],"contexts":["public_services"],"unknowns":["Whether the wrong-sex rejection came from an AI model, a rule engine or a data-matching step is not established.","The article gives the filing month as February without a year. The year 2025 is read from the article date (24 April 2025) and its present-tense context.","Dates and causes of the claimant's earlier sick-pay rejections are not stated, so this record covers only the February 2025 retirement claim.","No INSS decision record or independent confirmation of the claimant's account was inspected. The Portuguese rewrite adds no independent reporting.","Rest of World's statement that the app rejected requests from hundreds of people is an unsourced aggregate and is not counted."],"geography":{"basis":"The claimant is described as living in a remote town in northeast Brazil. Where the claim was filed from is not stated, so no event country is recorded. No court is involved.","court_countries":[],"event_countries":[],"affected_person_countries":["BR"]},"publication":{"basis":"A named-outlet journalist interviewed the claimant and the INSS, and the article reports a concrete rejected claim, its reported cause and its correction. The article gives the filing and approval months (February, March) without a year, so 2025 is inferred from its 24 April 2025 date. The AI role is reported and unverified, so claims are attributed and the record is marked alleged.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"Rest of World describes Meu INSS as an AI-powered app (computer vision and natural language processing over uploaded documents) and reports that the claim was rejected because the system identified the claimant as a man. The article does not say whether a model or a fixed rule produced this rejection, and the INSS describes the decisions as automated.","status":"reported"},"person_relations":["made_decision_about"]},"name":"Brazil: rural worker's retirement claim rejected through the Meu INSS app after the system reportedly recorded a wrong sex","summary":"In February (2025, inferred from the article date) a retirement claim filed through Brazil's Meu INSS social security app for a former sugarcane worker in northeast Brazil was rejected. Rest of World reported that the request was turned down because the system identified the claimant as a man. The claimant, who has chronic illnesses, is quoted saying the benefit is still denied despite complete documents, and Rest of World reports that minor errors in earlier claims through the app had led to numerous rejections. The retirement claim was approved in March after INSS directors corrected the mistake, which the claimant attributed to a contact at an agricultural workers' confederation. The INSS states that automated decisions follow specified legal criteria.","incidentDate":"2025-02-01","incidentKind":"single_event","incidentDatePrecision":"month","exposurePattern":"single_interaction","reportedDate":"2025-04-24","aiSystem":"Meu INSS app (AI-assisted automated analysis of benefit claims)","aiProduct":"Meu INSS app","aiCompany":"Dataprev (developer of the app, per Rest of World); INSS deploys it","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["other_material_harm"],"harmOutcomeSummary":"According to Rest of World, a retirement claim was rejected through the app, reportedly because the system recorded the wrong sex, and was approved in March after INSS directors corrected the error.","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"documented_minimum","affectedCountEvidence":"One claimant with a rejected and later corrected retirement claim. Rest of World says the app has rejected requests from hundreds of people for minor errors, which is an unsourced aggregate and is not counted.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://restofworld.org/2025/brazil-ai-social-security-app-rejected/","primarySourceLabel":"Rest of World, 'Brazil's AI-powered social security app is wrongly rejecting claims', 24 April 2025","firstPublishedAt":"2026-09-29T12:42:06.218216+00:00","updatedAt":"2026-09-30T01:16:57.309447+00:00","scopeVersion":"facts-v3","tags":["historical-2025"]},{"id":"2025-brazil-consul-ad-ted-talk-nc-state-senator-alleged-ai-alteration","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"The manipulated version of Plaintiff’s TED Talk changed her voice and visuals without her permission","relation":"supports","source_id":"s1"},{"locator":"fabricated statements and visuals to appear as though Plaintiff made them.","relation":"supports","source_id":"s1"},{"locator":"shocked to see myself saying things I never said.","relation":"supports","source_id":"s3"},{"locator":"The ad artificially manipulated CNN’s Brazil broadcasts and Salvador’s TED Talk, she said in the suit","relation":"supports","source_id":"s2"}],"assertion":"The plaintiff alleges that an advertising campaign for the Consul appliance brand used her 2018 TED Talk in altered form, with changed voice and visuals, so that she appeared to say things she did not say.","causal_attribution":"Allegation of the plaintiff, made in a complaint and in one interview. No defendant response is inspected. The complaint states the alteration was made \"using AI, or some other means\"."},{"id":"c2","status":"documented","evidence":[{"locator":"This follows the discovery that AI-generated and manipulated content was used within the case film to simulate real world events and campaign outcomes","relation":"supports","source_id":"s4"},{"locator":"we concluded that the only appropriate course of action was to mutually withdraw the entry from the Festival, and therefore withdraw the Grand Prix and associated prizes.","relation":"supports","source_id":"s4"}],"assertion":"Cannes Lions stated that AI-generated and manipulated content was used within the case film for the Consul campaign \"Efficient Way to Pay\" to simulate real world events and campaign outcomes, and withdrew the Grand Prix and associated prizes.","causal_attribution":"The statement establishes the festival finding and the withdrawal. It does not name the plaintiff or state which footage was altered."},{"id":"c3","status":"reported","evidence":[{"locator":"caused her reputational harm and severe emotional distress.","relation":"supports","source_id":"s1"},{"locator":"have subjected her to ridicule and embarrassment.","relation":"supports","source_id":"s1"},{"locator":"subjected her to ridicule and embarrassment,","relation":"supports","source_id":"s3"}],"assertion":"The plaintiff alleges that the campaign caused her reputational harm, ridicule and severe emotional distress.","causal_attribution":"Allegation in the complaint, repeated by two outlets. The complaint states the distress followed her discovery of the altered footage. Distress is not independently verified."},{"id":"c4","status":"documented","evidence":[{"locator":"CIVIL ACTION NO. 3:25-cv-628","relation":"supports","source_id":"s1"},{"locator":"Filed 08/20/25","relation":"supports","source_id":"s1"},{"locator":"OMNICOM GROUP, INC., WHIRLPOOL CORPORATION, DDB WORLDWIDE COMMUNICATIONS GROUP LLC, DM9DDB, WHIRLPOOL DO BRASIL LTDA.","relation":"supports","source_id":"s1"},{"locator":"Count I – Right of Publicity / Misappropriation of Likeness","relation":"supports","source_id":"s1"},{"locator":"Count II - Unfair and Deceptive Trade Practices","relation":"supports","source_id":"s1"},{"locator":"Count III - False Advertising, Endorsement and Light (Lanham Act","relation":"supports","source_id":"s1"},{"locator":"Count IV- Intentional Infliction of Emotional Distress","relation":"supports","source_id":"s1"},{"locator":"Count V - Defamation","relation":"supports","source_id":"s1"},{"locator":"misappropriated her likeness, used unfair and deceptive trade practices and violated the Lanham Act through false advertising","relation":"context","source_id":"s2"}],"assertion":"The plaintiff filed a federal civil complaint (W.D.N.C., case 3:25-cv-628) on 20 August 2025 against Omnicom Group, Whirlpool Corporation, DDB Worldwide Communications Group, DM9DDB and Whirlpool do Brasil (Consul), asserting five counts including misappropriation of likeness, unfair and deceptive trade practices, false advertising, intentional infliction of emotional distress and defamation.","causal_attribution":"The filing establishes its own existence and contents. Its allegations remain allegations."},{"id":"c5","status":"reported","evidence":[{"locator":"DM9 acknowledged “a series of errors” in the production and submission of the case video.","relation":"supports","source_id":"s5"},{"locator":"has stepped down in the wake of controversy surrounding its Cannes Lions Grand Prix-winning campaign","relation":"supports","source_id":"s5"},{"locator":"announced the immediate removal of chief creative officer","relation":"supports","source_id":"s2"}],"assertion":"DM9 publicly acknowledged \"a series of errors\" in producing and submitting the case video, and its chief creative officer was removed or stepped down.","causal_attribution":"Courthouse News relays a DM9 LinkedIn statement announcing the removal; Campaign Brief Asia relays Adweek for the step-down and DM9 for \"a series of errors\". Both are secondhand. The text inspected does not tie the errors to the plaintiff's footage specifically."},{"id":"c6","status":"reported","evidence":[{"locator":"had been altered using AI, or some other means","relation":"supports","source_id":"s1"}],"assertion":"The complaint alleges the plaintiff's name, image, likeness and persona were altered using AI, or some other means.","causal_attribution":"Plaintiff's allegation. The complaint does not identify the tool."}],"effects":[{"label":"Alleged reputational harm and emotional distress from altered depiction in an advertisement","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.courthousenews.com/wp-content/uploads/2025/08/deandrea-salvador-whirlpool-complaint.pdf","kind":"court_filing","access":"read","language":"en","translation_note":"","independence_group":"salvador-complaint"},{"id":"s2","url":"https://www.courthousenews.com/north-carolina-state-senator-sues-over-ai-manipulated-ad/","kind":"news","access":"read","language":"en","translation_note":"","independence_group":"salvador-complaint"},{"id":"s3","url":"https://www.route-fifty.com/artificial-intelligence/2025/09/north-carolina-lawmaker-sues-over-ai-generated-ad-campaign/407945/","kind":"news","access":"read","language":"en","translation_note":"","independence_group":"salvador-complaint"},{"id":"s4","url":"https://www.canneslions.com/news/cannes-lions-statement-dm9-entries-into-cannes-lions-2025","kind":"organisation_statement","access":"read","language":"en","translation_note":"","independence_group":"cannes-lions-statement"},{"id":"s5","url":"https://campaignbriefasia.com/2025/06/29/dm9-sao-paulo-co-president-and-cco-icaro-doria-steps-down-following-cannes-lions-controversy/","kind":"trade_press","access":"read","language":"en","translation_note":"","independence_group":"cannes-lions-statement"}],"version":1,"ai_roles":["others_use"],"contexts":["advertising","justice"],"unknowns":["Date the advertisement was created and first released is not established in inspected sources. The June 2025 event month rests on the Cannes award (19 June 2025) and the journalist contact (20 June 2025) in the complaint, so an earlier release, including before 2025, is not excluded.","Which AI tools altered the footage, and whether the alteration of the plaintiff's footage (as opposed to CNN Brasil footage) is the AI content found by Cannes Lions, is not established by the Cannes Lions statement.","The defendants' response to the complaint and the lawsuit outcome are not inspected.","The Washington Post article listed by the archive (30 November 2025) could not be retrieved (HTTP 403 via relay, no body) and is not cited.","The complaint's claim that the plaintiff's footage was altered was not independently verified against the advertisement itself."],"geography":{"basis":"Courthouse News describes the advertisement as promoting appliances in Brazil for Consul. The complaint states the plaintiff is a resident of North Carolina, United States, and was filed in the Western District of North Carolina.","court_countries":["US"],"event_countries":["BR"],"affected_person_countries":["US"]},"publication":{"basis":"Complaint text, a festival statement and press coverage inspected. Alteration and harm are the plaintiff's allegations. The festival finding of AI-manipulated content is documented in its own statement but does not name the plaintiff. Public figure named as the litigant in a public filing. The event month (June 2025) is the month of the Cannes Grand Prix (19 June) and of the journalist's contact with the plaintiff (20 June), the earliest dated events in the complaint (paras 52 and 56); the campaign's own release date is not stated and may precede June 2025.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"Claim c6 (complaint: AI, or some other means) and claim c2 (festival finding). The Cannes Lions statement records a festival finding of AI-generated and manipulated content in the case film. The complaint attributes the alteration of the plaintiff's footage to AI \"or some other means\". The plaintiff's footage is tied to AI only by her allegations and press reports of them.","status":"reported"},"person_relations":["depicted_or_impersonated"]},"name":"North Carolina state senator alleges an appliance advertisement used an altered version of her TED Talk, by AI or other means","summary":"State Senator DeAndrea Salvador alleges in a federal complaint filed in August 2025 that an advertisement for Whirlpool's Brazilian brand Consul, made by the agency DM9, used her 2018 TED Talk with altered voice and visuals so that she appeared to say things she did not say, and that this caused reputational harm and emotional distress. Cannes Lions stated on 27 June 2025 that AI-generated and manipulated content was used within the campaign's case film to simulate real-world events and campaign outcomes, and withdrew the Grand Prix. The festival statement does not name the plaintiff. The defendants' response to the complaint is not inspected.","incidentDate":"2025-06-01","incidentKind":"single_event","incidentDatePrecision":"month","exposurePattern":"unknown","reportedDate":"2025-06-27","aiSystem":"Unidentified AI tools: the complaint alleges alteration \"using AI, or some other means\", and Cannes Lions found AI-generated and manipulated content in the campaign case film; the tool and whether it was used on the plaintiff's footage are not established","aiProduct":"Unidentified AI tool","severity":"low","verificationStatus":"unverified","harmCategories":[],"harmOutcomes":["reputational_harm","psychological_distress"],"harmOutcomeSummary":"The plaintiff alleges in her complaint that the altered advertisement caused her reputational harm, ridicule and severe emotional distress.","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1,"affectedCountStatus":"documented_minimum","affectedCountEvidence":"One person, the plaintiff, who reports harm from the altered advertisement. CNN Brasil (an organisation), the festival juries and the public are not counted.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://www.courthousenews.com/north-carolina-state-senator-sues-over-ai-manipulated-ad/","primarySourceLabel":"Courthouse News Service, 21 August 2025: North Carolina state senator sues over AI-manipulated ad","firstPublishedAt":"2026-09-29T12:41:56.467139+00:00","updatedAt":"2026-09-30T01:16:57.505629+00:00","scopeVersion":"facts-v3","tags":["historical-2025"]}]}