{"meta":{"exportedAt":"2026-10-27T06:49:12.551Z","formatVersion":2,"selection":{"q":"pay","system":"","harm":"","context":"","country":"","role":"","relation":"core","evidence":"","year":"2022","response":"","severity":"","verification":"","view":"incidents","sort":"added"},"totalIncidents":3,"coverage":{"cases":3,"countries":2,"languages":2,"unknownLocation":1,"locationPending":0,"unknownLanguage":0,"unknownDate":0,"lawsuits":0,"regulatory":1,"minors":0,"coreRelations":3,"contextualRelations":0,"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":"2022-australia-federal-court-civil-penalty-for-ai-generated-deepfake-intimate-images-of-six-women","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'the respondent posted 12 non-consensual deepfake intimate images of six persons ordinarily resident in Australia online'; 'On 15 November 2022, [the respondent] uploaded to [the named website] a moving visual image'; 'On 27 May 2023, [the respondent] uploaded to [the named website] a moving visual image'; 'did not receive (and was aware that he had not received) consent from any of Depicted Persons 1, 2, 3, 4, 5 or 6'","relation":"supports","source_id":"s1"},{"locator":"'he posted deepfake nude images of six different women to a website'","relation":"supports","source_id":"s2"}],"assertion":"Between 15 November 2022 and 27 May 2023 the respondent posted 12 non-consensual deepfake intimate images, described as moving visual images, of six people ordinarily resident in Australia to a website accessible worldwide, without their consent.","causal_attribution":"Agreed facts accepted by the court as true; ABC reports the same finding. The judgment introduction gives 19 May 2023 as the end date, while the agreed facts and penalty orders include 27 May 2023."},{"id":"c2","status":"documented","evidence":[{"locator":"'created each of the deepfake images described below by following the same process: a. locating an image of the depicted person online; b. uploading the image to a webpage that allowed him to create semi-naked or naked images of the person; c. saving the generated image'","relation":"supports","source_id":"s1"}],"assertion":"He made each image by locating a photo of the person online, uploading it to a webpage that allowed him to create semi-naked or naked images of the person, and saving the generated image.","causal_attribution":"Agreed facts in the judgment; the tool is not named and its technique is not described."},{"id":"c3","status":"reported","evidence":[{"locator":"'this cruel, destructive and anti-social misuse of easily accessible AI tools'; 'for making a series of AI-generated deepfakes of prominent women'","relation":"supports","source_id":"s4"}],"assertion":"The eSafety Commissioner called this kind of deepfake abuse a misuse of easily accessible AI tools, and Information Age reported the images as AI-generated deepfakes.","causal_attribution":"The regulator that investigated the case, quoted by Information Age; the technical basis for the AI description is not stated beyond the agreed generation steps (c2)."},{"id":"c4","status":"documented","evidence":[{"locator":"'each of the Depicted Persons ordinarily resides in Australia and made a complaint about the images'; 'I am absolutely horrified about being caught up in this. Even though it is clearly a fake video, I feel violated, vulnerable and completely without agency'; 'the contraventions of s 75 are likely to have caused the Depicted Persons’ hurt and distress'; 'likely to have exacerbated that hurt and distress for Depicted Person 1 and Depicted Person 2'","relation":"supports","source_id":"s1"},{"locator":"'Even though it is clearly a fake video, I feel violated, vulnerable and completely without agency'","relation":"supports","source_id":"s2"}],"assertion":"Each of the six complained to eSafety; one complainant said she was horrified and, although it was clearly a fake video, felt violated, vulnerable and without agency; the court found the posts were likely to have caused the women hurt and distress, exacerbated for two women whose images stayed online for almost a year.","causal_attribution":"The court's finding, based on the complaints, the respondent's admission and expert evidence; the women gave no evidence of the extent of their loss."},{"id":"c5","status":"documented","evidence":[{"locator":"'I am not a resident of Australia. The removal notice means nothing to me. Get an arrest warrant if you think you are right.'; 'Some of the images [the respondent] posted online were retaliatory against individuals who were involved in the issuing of the Removal Notice.'; 'From around 3 November 2023, [the respondent] cooperated with eSafety in relation to removing the images'","relation":"supports","source_id":"s1"},{"locator":"'it meant nothing to him as he was not an Australian resident'","relation":"supports","source_id":"s3"}],"assertion":"He refused eSafety's May 2023 removal notice and remedial direction, saying the notice meant nothing to him as he was not a resident of Australia; he admitted that some later images were retaliation against people involved in issuing the notice; the images were removed only after he gave eSafety his log-in details from about 3 November 2023, following his arrest on other matters.","causal_attribution":"Agreed facts in the judgment; AAP reports the same reply."},{"id":"c6","status":"documented","evidence":[{"locator":"'This comes to a total of $343,500.'; 'The Commissioner says that this is the first proceeding involving the imposition of a civil penalty under the Act.'","relation":"supports","source_id":"s1"},{"locator":"'to pay a $343,500 penalty plus costs'","relation":"supports","source_id":"s3"}],"assertion":"On 26 September 2025 the court ordered penalties totalling A$343,500 and part of the Commissioner's costs, in what the Commissioner said was the first proceeding to impose a civil penalty under the Online Safety Act.","causal_attribution":"Court orders in the judgment."},{"id":"c7","status":"reported","evidence":[{"locator":"'was fined $25,000 for contempt of court, after admitting he breached three court orders made as part of the case'; 'he forwarded the documents to eSafety Commission staff and 49 other email addresses, including media outlets'","relation":"supports","source_id":"s2"},{"locator":"'to not publish the orders as they contained the names of the women targeted'","relation":"supports","source_id":"s4"}],"assertion":"In December 2023 he was fined A$25,000 for contempt after admitting he breached three orders in the case, including by forwarding the orders, which identified the women, to eSafety staff and 49 other email addresses, including media outlets.","causal_attribution":"Reported by ABC and Information Age from the court record; the contempt judgment was not inspected."},{"id":"c8","status":"reported","evidence":[{"locator":"'was found guilty of 17 offences by a jury on Tuesday'; 'digitally altered photographs of several women and girls he found on public websites in a way that depicted them either entirely or partially nude'; 'has been sentenced to two years in jail, which he has already served'","relation":"supports","source_id":"s5"}],"assertion":"In a separate criminal trial, a Brisbane jury convicted the same man in September 2026 of 17 offences over photos of women and girls, including students and staff of a Brisbane private school, that he digitally altered to depict them nude and emailed to the school and others, and that he was sentenced to two years in jail, which he had already served.","causal_attribution":"ABC report of a different proceeding, given as context only; whether its victims overlap with the six women is not established."}],"effects":[{"label":"women depicted without consent in generated deepfake intimate images posted online, which the court found likely caused them hurt and distress","claim_id":"c4","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.judgments.fedcourt.gov.au/judgments/Judgments/fca/single/2025/2025fca1191","kind":"court_record","access":"read","language":"en","translation_note":"Read in English on 2026-10-23 through an Internet Archive capture (web.archive.org/web/2025id_/...) because the court site returned HTTP 403; no translation. Full judgment with the agreed facts.","independence_group":"federal-court-judgment-2025-fca-1191"},{"id":"s2","url":"https://www.abc.net.au/news/2025-09-26/qld-deepfake-pornography-federal-court-charge/105822448","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-23 (HTTP 200); no translation. ABC report of the hand-down, with the earlier contempt fine.","independence_group":"abc-report-of-judgment"},{"id":"s3","url":"https://www.theguardian.com/technology/2025/sep/26/man-fined-34000-for-deepfake-pornography-of-prominent-australian-women-in-first-of-its-kind-case","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-23 (HTTP 200); no translation. AAP copy published by The Guardian (the Sydney Morning Herald carries the same AAP text).","independence_group":"aap-report-of-judgment"},{"id":"s4","url":"https://ia.acs.org.au/article/2025/australian-man-fined--350k-for-making-ai-deepfakes.html","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-23 (HTTP 200); no translation. Information Age report, including the eSafety Commissioner's LinkedIn statement.","independence_group":"information-age-report"},{"id":"s5","url":"https://www.abc.net.au/news/107106098","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-23 (HTTP 200); no translation. ABC report of the separate 2026 criminal trial; the short URL resolves to the full article.","independence_group":"abc-report-of-criminal-trial"}],"version":1,"ai_roles":["others_use"],"contexts":["privacy","justice"],"unknowns":["Which web tool generated the images and how it worked.","How many people viewed the images while they were online.","Whether any of the six women are among the victims in the separate 2026 Brisbane criminal case.","The extent of each woman's loss beyond the hurt and distress the court found likely; the women gave no evidence on it."],"geography":{"basis":"The judgment says each depicted person ordinarily resides in Australia and the Federal Court of Australia decided the case. Where he was when he posted the images is not stated (he told eSafety he was not resident in Australia), and the website was accessible worldwide, so the event country is left unknown.","court_countries":["AU"],"event_countries":[],"affected_person_countries":["AU"]},"publication":{"basis":"Published as a core case (depicted_or_impersonated): the Federal Court found, on admitted facts, that a man posted 12 non-consensual deepfake intimate images of six women that he generated by uploading their photos to an image-creation webpage, found the posts likely caused them hurt and distress, and imposed A$343,500 in penalties. The court record documents the generation, posting, complaints and penalty; the AI description is the regulator's.","reviewed_on":"2026-10-23"},"ai_involvement":{"basis":"The agreed facts accepted by the Federal Court state that the respondent uploaded each woman's photo to a webpage that allowed him to create semi-naked or naked images of her and saved the generated image, then posted those images; the court found the posts likely caused the women hurt and distress. The images depicted the women (depicted_or_impersonated). The eSafety Commissioner, whose office investigated the case, called this kind of deepfake abuse a misuse of easily accessible AI tools. The tool is not named and its technique is not described.","status":"supported"},"person_relations":["depicted_or_impersonated"]},"name":"Australian Federal Court orders man to pay A$343,500 for posting AI-generated deepfake intimate images of six women, in eSafety's first civil penalty case","summary":"On 26 September 2025 the Federal Court of Australia ordered a man to pay civil penalties of A$343,500 plus costs, in what the eSafety Commissioner said was the first proceeding to impose a civil penalty under the Online Safety Act 2021. On facts he admitted, he posted 12 non-consensual deepfake intimate images of six women ordinarily resident in Australia to a deepfake website between November 2022 and May 2023. He made each one by uploading a photo of the woman found online to a webpage that created semi-naked or naked images and saving the generated image. He refused eSafety's removal notice, so two images stayed online for almost a year, and he admitted posting some later images in retaliation against people involved in issuing the notice. Each woman complained to eSafety, and the court found the posts were likely to have caused them hurt and distress. The eSafety Commissioner called this kind of deepfake abuse a misuse of easily accessible AI tools.","incidentDate":"2022-11-15","incidentEndDate":"2023-05-27","incidentKind":"bounded_series","incidentDatePrecision":"day","exposurePattern":"unknown","reportedDate":"2025-09-26","aiSystem":"Unidentified web tool to which the poster uploaded a photo of each woman and which generated semi-naked or naked images (agreed facts in the judgment); the eSafety Commissioner called this kind of deepfake abuse a misuse of easily accessible AI tools","aiProduct":"Unidentified video tool","severity":"high","verificationStatus":"verified","harmCategories":[],"harmOutcomes":["exploitation_or_abuse","psychological_distress"],"harmOutcomeSummary":"According to the Federal Court's findings on admitted facts, six women were depicted without consent in generated deepfake intimate images posted online, two for almost a year, and the posts were likely to have caused them hurt and distress; one complainant said she felt violated, vulnerable and without agency.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":0,"otherPeopleHarmedMin":6,"affectedCountStatus":"exact","affectedCountEvidence":"Six Depicted Persons, each of whom complained to eSafety (judgment, agreed facts and [11]). Exact 6. People who viewed the images are not counted.","victimAgeRange":"adult","jurisdiction":"AU","platformType":"other","outcomeType":"regulatory_action","outcomeStatus":"resolved","primarySourceUrl":"https://www.judgments.fedcourt.gov.au/judgments/Judgments/fca/single/2025/2025fca1191","primarySourceLabel":"Federal Court of Australia, [2025] FCA 1191 (eSafety Commissioner civil penalty case), judgment of 26 September 2025","firstPublishedAt":"2026-10-23T03:08:18.073387+00:00","updatedAt":"2026-10-23T03:08:18.073387+00:00","scopeVersion":"facts-v3","tags":["deepfake","intimate-image-abuse","civil-penalty","online-safety-act","esafety-commissioner","australia","depicted-or-impersonated"]},{"id":"2022-british-columbia-air-canada-website-chatbot-bereavement-fare-advice-tribunal-orders-compensation","caseFacts":{"claims":[{"id":"c1","status":"documented","evidence":[{"locator":"'they interacted with a support chatbot'; 'a chatbot is an automated system that provides information to a person using a website in response to that person'; 'If you need to travel immediately or have already travelled and would like to submit your ticket for a reduced bereavement rate, kindly do so within 90 days of the date your ticket was issued'","relation":"supports","source_id":"s1"}],"assertion":"The tribunal records that on 11 November 2022 the passenger interacted with a support chatbot on Air Canada's website about bereavement fares, and that the chatbot's response, in a screenshot in evidence, said a ticket could be submitted for a reduced bereavement rate within 90 days of issue by someone who had already travelled.","causal_attribution":"Documented by the tribunal decision; the screenshot was in evidence and the tribunal later found the representation inaccurate."},{"id":"c2","status":"documented","evidence":[{"locator":"'the bereavement policy does not apply to requests for bereavement consideration after travel has been completed'; 'booked a one-way flight from Vancouver to Toronto, departing on November 12, for $794.98'; 'relying on the same information, they booked a one-way flight from Toronto to Vancouver, departing on November 18, for $845.38'","relation":"supports","source_id":"s1"}],"assertion":"The tribunal records that the airline's bereavement webpage said the policy did not apply after travel was completed, and that relying on the chatbot's information the passenger booked one-way flights between Vancouver and Toronto for $794.98 and $845.38.","causal_attribution":"Findings of fact in the decision."},{"id":"c3","status":"documented","evidence":[{"locator":"'an Air Canada representative responded and admitted the chatbot had provided'; 'said Air Canada had noted the issue so it could update the chatbot'","relation":"supports","source_id":"s1"}],"assertion":"The tribunal records that on 8 February 2023 an Air Canada representative admitted the chatbot had provided misleading words and said the airline had noted the issue so it could update the chatbot.","causal_attribution":"Documented by the decision from the airline's email in evidence."},{"id":"c4","status":"documented","evidence":[{"locator":"'This is a remarkable submission.'; 'I find Air Canada did not take reasonable care to ensure its chatbot was accurate.'; 'would not have flown last-minute if they knew they would have to pay the full fare'; 'a total of $812.02, broken down as follows'; '$650.88 in damages'","relation":"supports","source_id":"s1"},{"locator":"'Air Canada has been ordered to pay compensation to a grieving grandchild who claimed they were misled into purchasing full-price flight tickets by an ill-informed chatbot.'","relation":"supports","source_id":"s2"},{"locator":"'has been ordered to pay compensation after its chatbot gave a customer inaccurate information'","relation":"supports","source_id":"s3"}],"assertion":"The tribunal found negligent misrepresentation, rejected the airline's suggestion that the chatbot was a separate legal entity responsible for its own actions, found that Air Canada did not take reasonable care to ensure its chatbot was accurate, accepted that the passenger relied on the chatbot and would not have flown last-minute at full fare, and ordered the airline to pay $812.02 including $650.88 in damages.","causal_attribution":"The tribunal's own findings and order; the news reports relay the decision."},{"id":"c5","status":"reported","evidence":[{"locator":"'Last March, Air Canada’s chief information officer Mel Crocker told the Globe and Mail that the airline had launched the chatbot as an AI “experiment.”'; 'Air Canada’s initial investment in customer service AI technology was much higher than the cost of continuing to pay workers to handle simple queries.'","relation":"supports","source_id":"s4"}],"assertion":"Ars Technica reports that Air Canada's chief information officer told the Globe and Mail in March 2023 that the airline had launched the chatbot as an AI experiment, and describes the airline's investment in customer service AI technology.","causal_attribution":"The airline's own description of the chatbot, relayed by Ars Technica from a Globe and Mail interview that was not inspected; it describes the chatbot in general and does not say how the system produced the answer at issue."}],"effects":[{"label":"full fares paid in reliance on the chatbot's inaccurate bereavement-fare answer; $650.88 in damages found","claim_id":"c4","direction":"negative"}],"sources":[{"id":"s1","url":"https://decisions.civilresolutionbc.ca/crt/crtd/en/item/525448/index.do","kind":"court_record","access":"read","language":"en","translation_note":"Read in English on 2026-10-12 through the tribunal's public decision page (iframe content); no translation was needed and the research agent (an AI) read the text directly.","independence_group":"bccrt-2024-149-decision"},{"id":"s2","url":"https://www.cbc.ca/news/canada/british-columbia/air-canada-chatbot-lawsuit-1.7116416","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-12; the report relays the tribunal decision and quotes it, so it shares the decision's independence group.","independence_group":"bccrt-2024-149-decision"},{"id":"s3","url":"https://www.theguardian.com/world/2024/feb/16/air-canada-chatbot-lawsuit","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-12; the report relays the tribunal decision, so it shares the decision's independence group.","independence_group":"bccrt-2024-149-decision"},{"id":"s4","url":"https://arstechnica.com/tech-policy/2024/02/air-canada-must-honor-refund-policy-invented-by-airlines-chatbot/","kind":"news_report","access":"read","language":"en","translation_note":"Read in English on 2026-10-12 by the second researcher (an AI) from the publisher page; no translation was needed. The article relays the tribunal decision and, separately, remarks Air Canada's chief information officer made to the Globe and Mail in March 2023; only those relayed remarks are cited here, and the Globe and Mail interview was not inspected.","independence_group":"globe-and-mail-2023-air-canada-cio-interview"}],"version":1,"ai_roles":["institutional_use"],"contexts":["finance","everyday_life","justice"],"unknowns":["What kind of AI system the chatbot was and how it produced this answer; the airline gave the tribunal no information about its nature, and its chief information officer's description of the chatbot as an AI experiment is known only through Ars Technica's relay of a Globe and Mail interview.","Whether Air Canada changed the chatbot after the case."],"geography":{"basis":"The decision records the booking of flights between Vancouver and Toronto through Air Canada's website after a death in Ontario, and the tribunal is the British Columbia Civil Resolution Tribunal; where the passenger lives is not used to locate anything.","court_countries":["CA"],"event_countries":["CA"],"affected_person_countries":[]},"publication":{"basis":"Published on the tribunal decision, which documents the website chatbot's inaccurate answer, the passenger's reasonable reliance on it, the resulting full-fare purchase and an order for damages. The causal finding is the tribunal's; the chatbot's description as AI rests on the airline's chief information officer's remarks as relayed by Ars Technica, and how it produced this answer is recorded as unknown.","reviewed_on":"2026-10-12"},"ai_involvement":{"basis":"The tribunal records the chatbot's own generated answer (a screenshot saying a reduced bereavement rate could be claimed within 90 days of ticket issue after travel) and finds that the passenger reasonably relied on it to buy full-fare tickets, that the airline did not take reasonable care to ensure the chatbot was accurate, and that the reliance caused the loss. The tribunal describes the chatbot as an automated system answering a user's prompts and notes that Air Canada gave it no information about the chatbot's nature. Ars Technica reports that Air Canada's chief information officer told the Globe and Mail in March 2023 that the airline had launched the chatbot as an AI experiment (c5); how the system produced this particular answer is not established.","status":"reported"},"person_relations":["communicated_with"]},"name":"Air Canada website chatbot wrongly told a bereaved passenger a fare reduction could be claimed after travel; BC tribunal ordered the airline to pay damages","summary":"On 11 November 2022, after a grandparent died in Ontario, a passenger asked the support chatbot on Air Canada's website about bereavement fares. The chatbot's response, in a screenshot later filed in evidence, said a ticket could be submitted for a reduced bereavement rate within 90 days of issue, including after travel, while the airline's bereavement webpage said the policy did not apply after travel was completed. Relying on that answer, the passenger bought full-fare one-way tickets between Vancouver and Toronto for $794.98 and $845.38 and applied for the reduction on 17 November. In February 2023 an Air Canada representative admitted the chatbot had provided misleading words. On 14 February 2024 the British Columbia Civil Resolution Tribunal found negligent misrepresentation, rejected the airline's argument that it could not be held liable for its chatbot, found it had not taken reasonable care to ensure the chatbot was accurate, and ordered it to pay $812.02, including $650.88 in damages.","incidentDate":"2022-11-11","incidentEndDate":"2022-11-16","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"single_interaction","reportedDate":"2024-02-14","aiSystem":"Support chatbot on Air Canada's website answering a question about bereavement fares; the airline's chief information officer described it as an AI experiment (per Ars Technica), and the airline gave the tribunal no information about its nature","aiProduct":"Air Canada chatbot","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["financial_loss"],"harmOutcomeSummary":"The tribunal found that the passenger relied on the Air Canada chatbot's inaccurate answer about bereavement fares, paid $1,630.36 in full fares, and was owed $650.88 in damages, which with interest and fees the airline was ordered to pay.","frameworkFacets":[],"causationStatus":"established","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"exact","affectedCountEvidence":"One person: the applicant who relied on the chatbot's answer when buying full-fare tickets and to whom the tribunal awarded damages. Exact 1.","victimAgeRange":"unknown","jurisdiction":"CA","platformType":"chatbot","outcomeStatus":"resolved","primarySourceUrl":"https://decisions.civilresolutionbc.ca/crt/crtd/en/item/525448/index.do","primarySourceLabel":"Moffatt v. Air Canada, 2024 BCCRT 149, BC Civil Resolution Tribunal, 14 February 2024","firstPublishedAt":"2026-10-12T03:22:10.337605+00:00","updatedAt":"2026-10-12T03:22:10.337605+00:00","scopeVersion":"facts-v3","tags":["customer-service-chatbot","airline","wrong-information","tribunal-decision","negligent-misrepresentation","canada"]},{"id":"2022-taiwan-ai-voice-changer-romance-fraud-ring-20000-victims-nt900m-57-indicted","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"'更導入AI系統，將男員工、大媽的聲音變成甜美女生，在各大交友平台鎖定多金對象，談起網戀，要求對方送高價商品'; '將集團內22名實際從事「公關小姐」的甜美聲紋導入聊天系統中。因此不論男女，只要利用該系統聊天，出來的聲音都變成甜美女聲'","relation":"supports","source_id":"s1"},{"locator":"'警方表示，詐團還有工程師設計「AI變聲軟體」，偽裝成公關小姐要求被害人贈送禮物，甚至索討零用金、搬家費、買美容課程等款項'","relation":"supports","source_id":"s2"}],"assertion":"From 2022 a romance-fraud ring used AI voice-changing software, built by an engineer in the ring with the voiceprints of 22 hostesses, so that members chatting with targets on dating platforms sounded like young women.","causal_attribution":"Prosecutors' and police account relayed by two outlets (one chain); the software's role is their description and no court has ruled."},{"id":"c2","status":"reported","evidence":[{"locator":"'不法所得超過9億元，被害人超過2萬人。台北地檢署今日偵結，起訴57名被告'; '許多被害人因此舉債度日'","relation":"supports","source_id":"s1"},{"locator":"'從2022年開始，被害人超過2萬人，詐騙金額超過9億元'","relation":"supports","source_id":"s2"}],"assertion":"Prosecutors and police put the ring's victims at more than 20,000 and its criminal proceeds at more than NT$900 million since 2022, and say many victims went into debt.","causal_attribution":"Totals stated by prosecutors and police for the ring as a whole; the reports do not say how many victims heard the AI-converted voice."},{"id":"c3","status":"reported","evidence":[{"locator":"'台北地檢署於今年5、7月指揮刑事警察局第四大隊第三隊展開搜索，查扣集團購入的法拉利、麥拉倫超跑，以及47支名表，更扣押高達3億元的不動產與帳戶，另外也帶回67台裝有「AI深偽程式」的犯罪電腦'; '首腦夫妻分別被求處25年、18年徒刑'","relation":"supports","source_id":"s1"},{"locator":"'檢方2日偵結，依違反《詐欺犯罪危害防制條例》等罪嫌，起訴57名被告'","relation":"supports","source_id":"s2"}],"assertion":"After searches in May and July 2026 that seized supercars, 47 watches, up to NT$300 million in property and accounts and 67 computers holding AI deepfake programs, the Taipei District Prosecutors Office indicted 57 defendants on 2 September 2026 and asked for 25- and 18-year sentences for the couple who led the ring.","causal_attribution":"Indictment as reported by two outlets; allegations, not a court finding."}],"effects":[{"label":"more than 20,000 people defrauded of over NT$900 million in fake online romances, many left in debt","claim_id":"c2","direction":"negative"}],"sources":[{"id":"s1","url":"https://news.ltn.com.tw/news/society/breakingnews/5560352","kind":"news_report","access":"read","language":"zh","translation_note":"Read live in Traditional Chinese on 2026-10-02 (Liberty Times staff reporters, Taipei, 2 September 2026); reports the prosecutors' indictment findings. Researcher translation.","independence_group":"taipei-prosecutors-ai-voice-romance-ring"},{"id":"s2","url":"https://news.pts.org.tw/article/825232","kind":"news_report","access":"read","language":"zh","translation_note":"Read live in Traditional Chinese on 2026-10-02 (PTS, 2 September 2026); reports police and prosecutors' findings and quotes a Criminal Investigation Bureau officer. Same official chain as s1. Researcher translation.","independence_group":"taipei-prosecutors-ai-voice-romance-ring"}],"version":1,"ai_roles":["others_use"],"contexts":["relationships","finance"],"unknowns":["How many of the more than 20,000 victims heard the AI-converted voice, and how the NT$900 million loss is divided among them.","Whether the voice-changing software was built on an existing product or model.","Where the victims live, and the court outcome."],"geography":{"basis":"The ring was investigated by Taiwan's Criminal Investigation Bureau and indicted by the Taipei District Prosecutors Office, with the case transferred to the Taipei District Court (Liberty Times; PTS). The reports do not state where the victims live.","court_countries":["TW"],"event_countries":["TW"],"affected_person_countries":[]},"publication":{"basis":"Published under the 2026-09-15 charter as a core case: a fraud ring spoke to targets through AI voice-changing software in fake romances that prosecutors say defrauded more than 20,000 people. The account is the Taipei District Prosecutors Office's indictment as reported by the Liberty Times and PTS. The accused are named in reports and not named here.","reviewed_on":"2026-10-02"},"ai_involvement":{"basis":"Prosecutors and police say an engineer in the ring built AI voice-changing software that turned members' voices into a young woman's voice in the chat system used with targets (Liberty Times; PTS). The converted voice was a channel through which targets were spoken to (communicated_with). No forensic detail or underlying product is reported.","status":"reported"},"person_relations":["communicated_with"]},"name":"Taiwan: a romance-fraud ring used AI voice-changing software so members sounded like young women on dating platforms; prosecutors say more than 20,000 people were defrauded of over NT$900 million and indicted 57 people in September 2026","summary":"According to the Taipei District Prosecutors Office and Taiwan's Criminal Investigation Bureau, as reported by the Liberty Times and Public Television Service (PTS) on 2 September 2026, a fraud ring set up in 2022 by a married couple ran fake romances on dating platforms. An engineer in the ring built AI voice-changing software that loaded the voiceprints of 22 women working for the ring as hostesses into its chat system, so that any member, male staff and older women included, spoke to targets in a sweet young woman's voice. Members pressed the people they courted to buy expensive health supplements and gifts and to pay pocket money, moving costs and beauty courses, and the ring sent hostesses to meet some targets, in some cases for sex, to keep them paying. Prosecutors put the victims at more than 20,000 and the proceeds at more than NT$900 million since 2022, and say many victims are still living in debt. After searches in May and July 2026, the Taipei District Prosecutors Office indicted 57 defendants on 2 September 2026 under the Fraud Crime Hazard Prevention Act and asked for sentences of 25 and 18 years for the couple.","incidentDate":"2022-01-01","incidentKind":"bounded_series","incidentDatePrecision":"year","exposurePattern":"repeated_interactions","reportedDate":"2026-09-02","aiSystem":"AI voice-changing software built by an engineer in the fraud ring, which converted members' voices into a young woman's voice using the voiceprints of 22 hostesses (per prosecutors and police); investigators also seized 67 computers described as holding AI deepfake programs","aiProduct":"Unidentified voice-cloning tool","severity":"high","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["financial_loss"],"harmOutcomeSummary":"Prosecutors say more than 20,000 people were defrauded of a combined total exceeding NT$900 million by a romance-fraud ring whose members spoke to targets through AI voice-changing software, and that many victims went into debt (Taipei District Prosecutors Office via the Liberty Times and PTS).","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":20000,"affectedCountStatus":"documented_minimum","affectedCountEvidence":"Prosecutors and police put the ring's victims at more than 20,000 (被害人超過2萬人; Liberty Times and PTS), counted as a documented minimum of 20,000 people defrauded. The figure covers the ring's victims since 2022; the reports do not say how many of them heard the AI-converted voice. The 22 hostesses whose voiceprints were used and the employees reportedly punished by the leader are ring members and are not counted.","victimAgeRange":"unknown","jurisdiction":"TW","platformType":"other","outcomeType":"criminal_charges","outcomeStatus":"ongoing","primarySourceUrl":"https://news.ltn.com.tw/news/society/breakingnews/5560352","primarySourceLabel":"Liberty Times, 2 September 2026: 詐欺集團利用AI騙感情 逾2萬人受害損失9億元 起訴57人","firstPublishedAt":"2026-10-02T03:35:49.482296+00:00","updatedAt":"2026-10-02T03:35:49.482296+00:00","scopeVersion":"facts-v3","tags":["voice-changing","romance-scam","dating-platforms","fraud-ring","taiwan","taipei","indictment","communicated-with"]}]}