{"meta":{"exportedAt":"2026-09-30T05:35:05.874Z","formatVersion":2,"selection":{"q":"meta","system":"","harm":"","context":"","country":"","role":"","relation":"unknown","evidence":"","year":"","response":"","severity":"","verification":"","view":"incidents","sort":"added"},"totalIncidents":7,"coverage":{"cases":7,"countries":3,"languages":2,"unknownLocation":2,"locationPending":0,"unknownLanguage":1,"unknownDate":2,"lawsuits":0,"regulatory":0,"minors":0,"coreRelations":0,"contextualRelations":0,"mixedRelations":0,"unknownRelations":7,"relationPending":3,"relationUnknown":4},"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-instagram-users-report-unblockable-meta-ai-profiles","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"Article body: paragraph reporting Instagram users' blocking, restriction and reporting attempts.","relation":"supports","source_id":"s1"}],"assertion":"Axios reports that Instagram users were unable to block, restrict or report Meta's AI character accounts.","causal_attribution":"The inspected reporting describes this account. Claims remain attributed and causation is alleged; multiple reports of the same victim or police/court account are one independence group."},{"id":"c2","status":"reported","evidence":[{"locator":"Article body: spokesperson statement about blocking bug and removal.","relation":"supports","source_id":"s1"}],"assertion":"Meta acknowledged a bug affecting users' ability to block the accounts and said it was removing them.","causal_attribution":"The inspected reporting describes this account. Claims remain attributed and causation is alleged; multiple reports of the same victim or police/court account are one independence group."},{"id":"c3","status":"reported","evidence":[{"locator":"CNN body: Meta experimental-account statement and Grandpa Brian interview.","relation":"supports","source_id":"s2"}],"assertion":"The profiles came from a 2023 experiment; reporter chats also contained fictionalised biographies rather than verified real-person histories.","causal_attribution":"The inspected reporting describes this account. Claims remain attributed and causation is alleged; multiple reports of the same victim or police/court account are one independence group."}],"effects":[{"label":"User-reported inability to block or restrict AI character accounts","claim_id":"c2","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.axios.com/2025/01/03/meta-liv-social-media-ai-profiles","kind":"journalistic","access":"read","language":"en","translation_note":"","independence_group":"axios-meta-ai-blocking-report"},{"id":"s2","url":"https://www.cnn.com/2025/01/03/business/meta-ai-accounts-instagram-facebook/index.html","kind":"journalistic","access":"read","language":"en","translation_note":"","independence_group":"cnn-meta-ai-profile-report"}],"version":1,"ai_roles":["institutional_use"],"contexts":["privacy","everyday_life"],"unknowns":["The first blocking-failure date and number or locations of affected users are unknown.","Whether every affected user had previously chatted with a profile is unknown, so the person relation remains unknown."],"geography":{"basis":"The inspected reports do not establish affected users' countries. The CNN reporter's New York setting does not locate all users.","court_countries":[],"event_countries":[],"affected_person_countries":[]},"publication":{"basis":"An inspected source describes a concrete person-level adverse experience connected to the specified AI use. Statements remain attributed, private identities are minimized, and unknown dates, counts and technical details are preserved.","reviewed_on":"2026-09-30"},"ai_involvement":{"basis":"The outlet and Meta's quoted response identify these as AI-character accounts and connect the adverse experience to their blocking functionality. No specific user's technical logs were inspected.","status":"reported"},"person_relations":["unknown"]},"name":"Instagram users report they could not block Meta's AI character profiles","summary":"Axios reported that Instagram users could not block, restrict or report Meta's AI character accounts. Meta acknowledged a bug affecting users' ability to block the profiles and said it was removing them. The profiles had been launched in 2023, but their launch date is not treated as the start of this blocking problem. Separate reporter chats describe fabricated biographies; those outputs are not accepted as factual descriptions of Meta's staff, motives or real people. The affected user count and first blocking-failure date are unknown.","incidentKind":"bounded_series","incidentDatePrecision":"unknown","exposurePattern":"unknown","reportedDate":"2025-01-03","aiSystem":"Meta AI character profiles","aiProduct":"Meta AI","aiCompany":"Meta","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["loss_of_autonomy"],"harmOutcomeSummary":"User-reported inability to block or restrict AI character accounts","frameworkFacets":[],"causationStatus":"alleged","participantUsersAffectedMin":0,"otherPeopleHarmedMin":0,"affectedCountStatus":"unquantified","affectedCountEvidence":"The source gives no number of people unable to block profiles. Numeric fields are zero with unquantified status. Twenty-eight profiles are not twenty-eight harmed people.","victimAgeRange":"unknown","platformType":"chatbot","primarySourceUrl":"https://www.axios.com/2025/01/03/meta-liv-social-media-ai-profiles","primarySourceLabel":"Axios, 3 January 2025, user reports and Meta response","firstPublishedAt":"2026-09-30T01:11:52.447069+00:00","updatedAt":"2026-09-30T01:24:52.150087+00:00","scopeVersion":"facts-v3","tags":[]},{"id":"2026-kenya-sama-meta-ai-annotation-contract-ended-over-1000-redundancies-announced","caseFacts":{"claims":[{"id":"c1","status":"reported","evidence":[{"locator":"announced on Thursday that the workers were being laid off after Meta terminated a contract.","relation":"supports","source_id":"s1"},{"locator":"More than 1,000 low-paid workers in Kenya have been abruptly sacked by an outsourcing company contracted by Meta","relation":"supports","source_id":"s1"},{"locator":"The sacked workers, many involved in AI training, have been given six days’ notice, according to the Oversight Lab","relation":"supports","source_id":"s1"},{"locator":"In a press release on the 16th April, Sama states it recently received a formal notice by Meta, a key client, that was ending its contract with the annotation firm","relation":"supports","source_id":"s2"},{"locator":"Less than two months later, Meta ended its contract with Sama, which Sama said would result in 1,108 workers being made redundant.","relation":"supports","source_id":"s3"},{"locator":"Sama claims that Meta’s cancellation of the contract affected 1,108 workers.","relation":"supports","source_id":"s4"},{"locator":"On April 16, that routine came to an end.","relation":"supports","source_id":"s5"},{"locator":"Sama’s April 16 press release announcing the mass layoffs spoke of compliance with Section 40 of Kenya’s Employment Act","relation":"supports","source_id":"s5"}],"assertion":"On 16 April 2026 Sama, an outsourcing company with operations in Nairobi that did AI-training and data-annotation work for Meta, announced that Meta had given formal notice ending its contract and that a redundancy process would affect more than 1,000 employees. Sama put the number at 1,108 workers. The Oversight Lab, an organisation that advocates for fair technology regulation in Africa, said the workers had six days' notice.","causal_attribution":"The announcement and the 'more than 1 000' figure trace to Sama's 16 April press release as reported by SvD, and 1,108 is Sama's figure as reported by BBC and Ars Technica, so they are one chain. The Guardian states 'more than 1,000' in its own voice without a stated source and attributes only the six days' notice to the Oversight Lab. No press release or redundancy notice was inspected directly, and the final number made redundant is not reported."},{"id":"c2","status":"reported","evidence":[{"locator":"Last month, we paused our work with Sama while we looked into these claims.","relation":"supports","source_id":"s2"},{"locator":"We’ve also decided to end our work with Sama because they don’t meet our standards.","relation":"supports","source_id":"s1"},{"locator":"Meta says it's because Sama did not meet its standards, a criticism Sama rejects.","relation":"supports","source_id":"s3"},{"locator":"At no point were we notified of any failure to meet those standards","relation":"supports","source_id":"s3"},{"locator":"After the investigation was published at the end of February, all projects at the subcontractor Sama in Kenya linked to Meta’s smart glasses were halted.","relation":"supports","source_id":"s2"}],"assertion":"Meta said in a statement reported on 17 April 2026 that 'last month' it paused its work with Sama while it looked into the claims, and that it decided to end its work with Sama because Sama does not meet its standards. Sama says it was never notified of any failure to meet those standards. SvD reports that after the investigation was published all Sama projects linked to Meta's smart glasses were halted.","causal_attribution":"Meta's and Sama's own statements to the outlets, each attributed. SvD says Meta declined to answer follow-up questions, and no source shows Meta's internal reasoning."},{"id":"c3","status":"disputed","evidence":[{"locator":"A Kenyan workers' organisation alleges Meta's decision was caused by the staff speaking out.","relation":"supports","source_id":"s3"},{"locator":"Meta has not addressed that allegation","relation":"context","source_id":"s3"},{"locator":"the company has tried to identify who spoke with journalists","relation":"supports","source_id":"s2"},{"locator":"“After the exposé, they started looking for people who could have leaked the information,”","relation":"supports","source_id":"s5"},{"locator":"“From what they told us, they said it was because of the exposé from the Swedish journalists,”","relation":"supports","source_id":"s5"},{"locator":"Meta had pulled a major contract citing only compliance with Kenyan employment law.","relation":"context","source_id":"s5"},{"locator":"we do not monitor or track employees for the purpose of identifying individuals who may have spoken to the media","relation":"contradicts","source_id":"s2"},{"locator":"we reject any suggestion of retaliatory behavior.","relation":"contradicts","source_id":"s2"}],"assertion":"A Kenyan workers' organisation alleges Meta's decision was caused by the staff speaking out, and, when the BBC reported it on 30 April 2026, Meta had not addressed that allegation. Employees told SvD that Sama tightened security and tried to identify who had spoken to journalists. A worker quoted by TechPolicy.Press says the workers were told the termination was because of the Swedish investigation, without saying who told them. Sama says it does not monitor or track employees to identify who spoke to the media and rejects any suggestion of retaliatory behaviour.","causal_attribution":"Allegations by a workers' organisation and by unnamed employees, denied or unaddressed by the companies. The TechPolicy.Press worker also says Meta cited only compliance with Kenyan employment law, which differs from Meta's public statement about standards. No document showing the reason for the termination was inspected."},{"id":"c4","status":"reported","evidence":[{"locator":"data annotation work, including working with video, image, and speech annotation for Meta’s AI systems for Ray-Ban Metas.","relation":"supports","source_id":"s4"},{"locator":"The workers the Swedish newspapers spoke to were data annotators, teaching Meta's AI to interpret images by manually labelling content.","relation":"supports","source_id":"s3"},{"locator":"The sacked workers, many involved in AI training, have been given six days’ notice, according to the Oversight Lab","relation":"supports","source_id":"s1"},{"locator":"Last month Meta paused its work with Sama after allegations about the workers viewing private scenes filmed using the company’s Ray-Ban smart glasses","relation":"supports","source_id":"s1"}],"assertion":"Sama's Meta work included video, image and speech annotation for Meta's AI systems for Ray-Ban Meta glasses. The Sama annotators who told the Swedish newspapers about viewing private footage from the glasses were data annotators of this kind, and the Guardian reports that many of the workers made redundant were involved in AI training. Meta paused its work with Sama after allegations about the workers viewing private scenes filmed with the glasses.","causal_attribution":"Outlet descriptions of the work. Which of the redundant workers worked on the glasses projects is not reported, and no source states that the workers made redundant were the same people who spoke to the newspapers."}],"effects":[{"label":"More than 1,000 Sama workers in Kenya, many involved in AI training, were given notice of redundancy after Meta ended its contract, on six days' notice according to the Oversight Lab","claim_id":"c1","direction":"negative"},{"label":"Employees report tightened security and efforts to identify who spoke to journalists after the investigation (Sama denies tracking employees to identify sources)","claim_id":"c3","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.theguardian.com/technology/2026/apr/17/kenyan-outsourcing-company-for-meta-sacks-workers","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"guardian-sama-oversight-lab"},{"id":"s2","url":"https://www.svd.se/a/zOlP6K/meta-halts-ai-training-after-svd-gp-investigation","kind":"news_report","access":"read","language":"en","translation_note":"SvD's own English edition of the article, read in English. The Swedish investigation of 25 February 2026 (not cited) was read in Swedish by a language model with no human review.","independence_group":"svd-gp-follow-up"},{"id":"s3","url":"https://www.bbc.com/news/articles/c5y7yvgy0w6o","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"bbc-sama-meta-statements"},{"id":"s4","url":"https://arstechnica.com/gadgets/2026/04/meta-cuts-contractors-who-reported-seeing-ray-ban-meta-users-have-sex/","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"bbc-sama-meta-statements"},{"id":"s5","url":"https://www.techpolicy.press/how-ais-labor-supply-chains-fail-workers/","kind":"news_report","access":"read","language":"en","translation_note":"","independence_group":"techpolicy-worker-interviews"}],"version":1,"ai_roles":["institutional_use"],"contexts":["work","privacy"],"unknowns":["Sama's press release and Meta's notice of termination were not inspected. The announcement is known through outlets and company statements.","The final number of workers made redundant and how many of them worked on the glasses projects are not reported. The Guardian says 'many' were involved in AI training.","The reason for the termination is disputed and no inspected source shows Meta's internal reasoning. Meta declined follow-up questions from SvD.","The retaliation account rests on a workers' organisation, unnamed employees and one worker quoted by TechPolicy.Press. That worker also says Meta cited only compliance with Kenyan employment law, which differs from Meta's public statement about standards.","No source states that the workers made redundant were the same people who spoke to the newspapers, and the annotators' accounts of the footage are anonymous interview accounts.","SvD's 19 April 2026 article was paywalled (two paragraphs read), The Standard's article on the Kenyan regulator was paywalled, and the SvD article in this record was read in its English edition.","Outlets differ on where Sama is headquartered (the Guardian says based in Nairobi, the BBC says US headquartered, TechPolicy.Press says San Francisco-based with offices in Nairobi, Ars Technica says Kenya-headquartered). This record uses only the location of the workers.","This record covers the April 2026 redundancies. The annotators' reported viewing of glasses users' footage started before 2026 and is not part of this record."],"geography":{"basis":"The Guardian describes the workers as 'in Kenya' and SvD refers to Sama's projects in Kenya and its Nairobi operations. Outlets differ on where Sama is headquartered and that is not used. Where Meta took the decision is not stated and is not used.","court_countries":[],"event_countries":["KE"],"affected_person_countries":["KE"]},"publication":{"basis":"The Guardian (17 April), SvD and GP (17 April), BBC (30 April), Ars Technica and TechPolicy.Press (12 June 2026) were read in full from the saved bodies. The redundancy fact and figure trace to Sama's announcement and are left at reported. Meta's stated reasons and Sama's replies are attributed, and the retaliation allegation is recorded as disputed. The workers are described by role and are not named.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"The workers were data annotators whose work trained Meta's AI systems for its Ray-Ban Meta glasses (c4), and Meta says it paused the Sama work while it looked into claims about annotators viewing private footage from the glasses (c2). No AI system decided anything about the workers: Meta ended the contract and Sama announced the redundancies. The AI link is that the redundancies followed, and are alleged by a workers' organisation to result from, annotators' accounts of a workplace AI practice. Meta gives a different reason and Sama says it was not told of any failure (c3). No relation between an AI system and the workers is established, so the relation recorded is unknown.","status":"disputed"},"person_relations":["unknown"]},"name":"Kenya: Sama announces more than 1,000 redundancies (1,108 by its own figure) after Meta ends its AI data-annotation contract, less than two months after annotators described intimate footage from Meta's AI glasses, and whether the two are linked is disputed","summary":"On 16 April 2026 Sama, an outsourcing company with operations in Nairobi that did AI data-annotation work for Meta, announced that Meta had given formal notice ending its contract and that a redundancy process would affect more than 1,000 employees. Sama put the figure at 1,108 workers, and the Oversight Lab said the workers had six days' notice. The decision came less than two months after the SvD and GP investigation in which Sama annotators said they had seen intimate footage from users of Meta's AI glasses. Meta said it paused its work with Sama the month before while it looked into those claims and then decided to end the work because Sama does not meet its standards. Sama says it was never notified of any failure to meet those standards. A Kenyan workers' organisation alleges Meta's decision was caused by staff speaking out, which Meta had not addressed when the BBC reported it on 30 April 2026. Employees told SvD that Sama tightened security and tried to identify who had spoken to journalists, which Sama denies. The workers are described by role and are not named.","incidentDate":"2026-04-16","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"unknown","reportedDate":"2026-04-17","aiSystem":"Human review of Meta AI assistant data from Ray-Ban Meta glasses by Sama annotators (AI training and data-annotation work), as reported","aiProduct":"Meta AI (reported)","aiCompany":"Meta","severity":"medium","verificationStatus":"unverified","harmCategories":[],"harmOutcomes":["professional_harm"],"harmOutcomeSummary":"Sama announced that more than 1,000 employees in Kenya, many involved in AI training for Meta, would be made redundant (1,108 by Sama's figure, on six days' notice according to the Oversight Lab) after Meta ended its contract. Whether the termination responded to annotators speaking about footage from Meta's AI glasses is disputed: a Kenyan workers' organisation alleges it did, Meta says Sama does not meet its standards, and Sama says it was never notified of a failure.","frameworkFacets":[],"causationStatus":"disputed","participantUsersAffectedMin":0,"otherPeopleHarmedMin":1000,"affectedCountStatus":"documented_minimum","affectedCountEvidence":"The Guardian states 'more than 1,000' workers were sacked without saying where the figure comes from, and SvD's 17 April article says Sama's press release announced a redundancy process that 'will affect more than 1 000 employees'. BBC and Ars Technica give Sama's own figure of 1,108. The lower bound of 1,000 is counted and the 1,108 figure is attributed to Sama. The Guardian says many of the workers were involved in AI training and does not say how many. Dependants are not counted. The final number made redundant is not reported.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://www.theguardian.com/technology/2026/apr/17/kenyan-outsourcing-company-for-meta-sacks-workers","primarySourceLabel":"The Guardian (17 Apr 2026): Kenyan firm sacks more than 1,000 workers after losing Meta contract","firstPublishedAt":"2026-09-29T21:16:44.033365+00:00","updatedAt":"2026-09-30T01:17:42.812683+00:00","scopeVersion":"facts-v3","tags":["historical-2026"]},{"id":"2026-meta-mci-us-employee-keystroke-and-screen-capture-for-ai-training-then-internal-data-exposure","caseFacts":{"claims":[{"id":"c1","status":"corroborated","evidence":[{"locator":"Meta is installing tracking software on U.S. employees’ work computers that will capture mouse movements, clicks, and keystrokes, along with some screenshots to feed the data into its AI training pipeline, according to Reuters.","relation":"supports","source_id":"s1"},{"locator":"will run on a designated list of work apps and websites.","relation":"supports","source_id":"s1"},{"locator":"The company added that safeguards are in place to protect sensitive content and that the data will not be used for any other purpose.","relation":"context","source_id":"s1"},{"locator":"allows Meta to observe and collect data from staffers' actions on their work computers","relation":"supports","source_id":"s3"},{"locator":"A Meta spokesperson confirmed the project","relation":"supports","source_id":"s3"},{"locator":"To help, we're launching an internal tool that will capture these kinds of inputs on certain applications to help us train our models.","relation":"supports","source_id":"s3"},{"locator":"It’s a piece of mandatory software that Meta began installing on the laptops of US employees last month.","relation":"supports","source_id":"s4"},{"locator":"The company positioned the program as an opportunity for Meta employees to “help our models get better simply by doing their daily work,” with a promise that the information would not be used for performance reviews or other potentially invasive purposes.","relation":"context","source_id":"s2"}],"assertion":"From April 2026 Meta installed software on US employees' work computers that captures mouse movements, clicks, keystrokes and screen content on a designated list of applications and sites, and Meta said the inputs are used to train its AI models.","causal_attribution":"Meta's own statement of purpose (training its models) is quoted by several outlets from the same company statement. Reuters' original memo report could not be read (HTTP 401 and no archive copy), so the Reuters chain is inspected only through Fortune and Gizmodo. CNBC and WIRED report from their own internal messages and sources."},{"id":"c2","status":"reported","evidence":[{"locator":"When MCI launched, employees couldn’t opt out, but that changed to a limited degree after workers protested.","relation":"supports","source_id":"s6"},{"locator":"It’s a piece of mandatory software that Meta began installing on the laptops of US employees last month.","relation":"supports","source_id":"s4"},{"locator":"Meta this month began offering more exemptions to the monitoring, including letting staffers briefly turn off the surveillance so they could complete sensitive tasks, such as scheduling a personal appointment","relation":"supports","source_id":"s5"},{"locator":"can control what shows up on your screen by not doing personal work on your work computer,\" the memo said.","relation":"context","source_id":"s3"}],"assertion":"WIRED reports that MCI was mandatory software for US employees and that at launch employees could not opt out. WIRED also reports, citing two people familiar with the matter, that in June 2026 Meta began offering more exemptions, including letting staffers briefly turn off the tracking for sensitive tasks.","causal_attribution":"WIRED reports the opt-out position from its own sources. The CNBC memo excerpt shows Meta's launch-time answer to employees' concerns."},{"id":"c3","status":"reported","evidence":[{"locator":"Multiple Meta employees characterized the data-tracking project as \"dystopian\" in internal messages viewed by CNBC.","relation":"supports","source_id":"s3"},{"locator":"Others expressed concerns that MCI could widely expose sensitive data, including user passwords, details about new product development, and personal information about workers' immigration status, health or family members.","relation":"supports","source_id":"s3"},{"locator":"“Selfishly, I don't want my screen scraped because it feels like an invasion of my privacy,” wrote an engineer in an internal post seen by nearly 20,000 coworkers this week.","relation":"supports","source_id":"s4"}],"assertion":"Employees objected internally: CNBC reports multiple employees called the project \"dystopian\" in internal messages and others worried it could expose sensitive data such as passwords and personal information, and WIRED quotes an engineer's internal post, seen by nearly 20,000 coworkers, saying having the screen scraped felt like an invasion of privacy.","causal_attribution":"Employees' own stated concerns as quoted by CNBC and WIRED from internal messages. These are reported reactions and concerns, and neither outlet reports that the concerns had materialised at that point."},{"id":"c4","status":"reported","evidence":[{"locator":"Meta employees at several US offices walked into meeting rooms, broke for coffee at vending machines, and used the restrooms only to find pamphlets denouncing the company’s new mouse-tracking software as an “Employee Data Extraction Factory” and urging staff to sign an online petition against it.","relation":"supports","source_id":"s7"},{"locator":"In Meta offices in California and New York, workers have been posting flyers in cafeterias and other communal areas pointing colleagues to the petition.","relation":"supports","source_id":"s4"},{"locator":"Last month, more than 1,600 employees at the tech giant signed an internal petition protesting the laptop surveillance effort","relation":"supports","source_id":"s5"}],"assertion":"In May 2026 flyers appeared in US offices pointing staff to a petition against the programme, and WIRED reported in June that more than 1,600 employees had signed an internal petition protesting it.","causal_attribution":"The TNW piece relays Reuters reporting on the flyers. WIRED reports the flyers and the signature count from its own sources."},{"id":"c5","status":"reported","evidence":[{"locator":"Meta left potentially sensitive information collected from employee laptops accessible to anyone inside the company, according to an internal security notice seen by WIRED and three current employees familiar with the issue.","relation":"supports","source_id":"s5"},{"locator":"The security notice sent out Monday indicated that “employee data across 45,000 hive tables,” had been exposed.","relation":"supports","source_id":"s5"},{"locator":"Those tables included employee activity such as “full prompts and transcriptions, private conversations, people and performance data,” according to documents viewed by WIRED.","relation":"supports","source_id":"s5"},{"locator":"we have no indication at this time that any data was improperly accessed by Meta employees, we're pausing it while we investigate,","relation":"supports","source_id":"s5"},{"locator":"The issue made “some MCI-derived data” accessible to more people than intended","relation":"supports","source_id":"s6"}],"assertion":"According to WIRED, an internal security notice sent on 22 June 2026 said employee data across 45,000 hive tables had been exposed to anyone inside Meta, including \"full prompts and transcriptions, private conversations, people and performance data\". Meta said it had no indication the data was improperly accessed and paused the programme.","causal_attribution":"WIRED reports the exposure from an internal security notice and current employees, and a Meta vice president's note to staff confirmed that some MCI-derived data was accessible to more people than intended. WIRED says the data \"is believed to include\" keystrokes, mouse clicks and screen content and that it was not immediately clear whether AI played a role in the access-control failure. No source reports that anyone misused the data."},{"id":"c6","status":"reported","evidence":[{"locator":"said that the tracking program’s implementation had fallen short of the standards outlined in its privacy review","relation":"supports","source_id":"s5"},{"locator":"the security issue had been discovered on June 18 and addressed within four hours. But the initial fix didn’t stick and access to the data had to be further locked down.","relation":"supports","source_id":"s6"},{"locator":"“We will only re-enable MCI when we are confident in the effectiveness of our data protection controls,”","relation":"supports","source_id":"s6"}],"assertion":"WIRED reports that Meta's chief technology officer said in an internal post that the programme's implementation had fallen short of the standards in its privacy review, and that a Meta vice president told staff the issue was discovered on 18 June and addressed within four hours, that the initial fix did not hold and access had to be locked down further, and that MCI would be re-enabled only when Meta was confident in its data protection controls.","causal_attribution":"Internal statements by Meta executives as reported by WIRED from posts it saw. They are the company's own account of its conduct and stay at reported."}],"effects":[{"label":"US employees required to run software recording their keystrokes, mouse activity and screen content for AI training, initially with no opt-out","claim_id":"c2","direction":"negative"},{"label":"Captured employee activity data left accessible to anyone inside the company (discovered 18 June 2026), with Meta reporting no indication of improper access","claim_id":"c5","direction":"negative"}],"sources":[{"id":"s1","url":"https://fortune.com/2026/04/21/meta-will-start-tracking-employees-screens-and-keystrokes-to-train-ai/","kind":"news_relay","access":"read","language":"en","translation_note":"","independence_group":"reuters-mci-memo-2026-04-21"},{"id":"s2","url":"https://gizmodo.com/meta-plans-to-turn-its-employees-clicks-and-keystrokes-into-ai-training-data-2000749176","kind":"news_relay","access":"read","language":"en","translation_note":"","independence_group":"reuters-mci-memo-2026-04-21"},{"id":"s3","url":"https://www.cnbc.com/2026/04/22/meta-tracks-employee-usage-on-google-linkedin-ai-training-project.html","kind":"news_report","access":"read","language":"en","translation_note":"Read from an Internet Archive capture (23 Apr 2026) because the live page returned HTTP 403.","independence_group":"cnbc-mci-internal-messages"},{"id":"s4","url":"https://www.wired.com/story/meta-employee-protest-mouse-tracking-surveillance-ai-training/","kind":"original_news_reporting","access":"read","language":"en","translation_note":"","independence_group":"wired-mci-reporting"},{"id":"s5","url":"https://www.wired.com/story/meta-accidentally-let-employees-access-each-others-keystroke-data/","kind":"original_news_reporting","access":"read","language":"en","translation_note":"","independence_group":"wired-mci-reporting"},{"id":"s6","url":"https://www.wired.com/story/meta-pauses-employee-tracking-program-following-internal-security-breach/","kind":"original_news_reporting","access":"read","language":"en","translation_note":"","independence_group":"wired-mci-reporting"},{"id":"s7","url":"https://thenextweb.com/news/meta-mouse-tracking-protest-layoffs","kind":"news_relay","access":"read","language":"en","translation_note":"","independence_group":"reuters-mci-protest-2026-05-12"}],"version":1,"ai_roles":["institutional_use"],"contexts":["work","privacy"],"unknowns":["Reuters' original reports (21 April memo, 12 May protest, 22 June exposure) could not be read (HTTP 401 direct, and every Internet Archive capture tried for April, May and June 2026 was an anti-bot stub). Reuters detail is taken only from the Fortune, Gizmodo and The Next Web relays.","No source states how many employees' data was in the exposed tables or how many employees are covered by MCI.","No inspected source states when the access-control misconfiguration began or how long the data was accessible. Meta's vice president said the issue was discovered on 18 June 2026 and that the first fix did not hold, and WIRED quotes the chief technology officer as saying the setup had misconfigured access control lists.","A separate complaint, Does 1 through 26 v. Meta Platforms (N.D. Cal., filed 13 July 2026), is covered in the record 2026-us-meta-26-employees-sue-alleging-ai-assisted-may-layoff-selection-penalized-protected-leave. It alleges on information and belief that keystroke and screen-content monitoring data supplied inputs to AI-assisted layoff scoring, and Meta denies that AI was used in the selection. That allegation is not part of this record, and no source inspected for this record reports a job consequence for an employee from MCI data.","Whether anyone accessed or misused the exposed data is unknown. Meta said it had no indication of improper access at the time of the pause.","Some aggregator articles say the exposed data included tax and medical records. An employee's comment cited by Reuters (via eWeek) only said personal tax and medical information is accessible through work computers, and eWeek says there was no confirmation those details were in the exposed records, so it is not recorded.","WIRED's report that Meta removed some protest posters rests on two anonymous employees and Meta declined to comment, so it is not recorded as a fact.","Whether the pause is permanent or MCI later resumed was not established beyond the 22 June 2026 announcement."],"geography":{"basis":"Fortune and WIRED report the tracking running on US employees' work computers and WIRED says only US employees are currently subject to it. The country is not taken from the company's headquarters.","court_countries":[],"event_countries":["US"],"affected_person_countries":["US"]},"publication":{"basis":"Fortune, Gizmodo, CNBC (Internet Archive capture), The Next Web and three WIRED articles were read in full. Meta's statements and its executives' internal posts are the company's own account of its conduct and stay at reported. The launch and the exposure rest on separate reporting chains (Reuters via relays, CNBC's own messages, WIRED's own sources) but the exposure is inspected only through WIRED and one Meta executive's note quoted by WIRED. No employee is named.","reviewed_on":"2026-09-29"},"ai_involvement":{"basis":"A Meta spokesperson said the tool captures inputs on certain applications to help train Meta's models, which is the company's own account of the AI purpose. The AI role is as the reason for the data collection. No inspected source describes an AI output, decision or action affecting an employee, and WIRED says it was not immediately clear whether AI played a role in the access-control failure.","status":"reported"},"person_relations":["unknown"]},"name":"US: Meta records employees' keystrokes and screens to train AI agents, then leaves some of the captured data accessible company-wide and pauses the programme","summary":"From April 2026 Meta installed its Model Capability Initiative (MCI) on US employees' work computers, recording mouse movements, clicks, keystrokes and screen content on designated apps and sites so that its AI agents could learn how people use software. Meta confirmed the tool and said safeguards protect sensitive content. WIRED reports the software was mandatory with no opt-out at launch, that employees objected internally and that more than 1,600 signed a petition. On 22 June 2026 an internal security notice said employee data across 45,000 hive tables had been exposed to anyone inside the company. Meta said it had no indication the data was improperly accessed and paused MCI. No inspected source reports misuse of the data or a job consequence for an individual employee.","incidentDate":"2026-04-01","incidentEndDate":"2026-06-22","incidentKind":"bounded_series","incidentDatePrecision":"range","exposurePattern":"unknown","reportedDate":"2026-04-21","aiSystem":"Model Capability Initiative (MCI): Meta's internal software capturing employee computer inputs and screen content to train AI agents","aiProduct":"Meta Model Capability Initiative","aiCompany":"Meta Platforms","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["loss_of_autonomy","other_material_harm"],"harmOutcomeSummary":"WIRED reports that Meta made the recording software mandatory for US employees, with no opt-out at launch, and that in June 2026 an internal notice said employee data across 45,000 hive tables was exposed to anyone inside the company. Meta said it had no indication the data was improperly accessed. No inspected source reports misuse of the data or a consequence for an individual employee.","frameworkFacets":[],"causationStatus":"unclear","participantUsersAffectedMin":0,"otherPeopleHarmedMin":0,"affectedCountStatus":"unquantified","affectedCountEvidence":"No inspected source counts employees harmed. The programme covers US employees, the exposure notice gives no number of affected employees, and the 1,600 petition signers are not counted as harmed people. Numeric zeros are placeholders.","victimAgeRange":"adult","platformType":"other","primarySourceUrl":"https://www.wired.com/story/meta-accidentally-let-employees-access-each-others-keystroke-data/","primarySourceLabel":"WIRED (22 Jun 2026)","firstPublishedAt":"2026-09-29T21:16:02.93024+00:00","updatedAt":"2026-09-30T01:17:45.697526+00:00","scopeVersion":"facts-v3","tags":["historical-2026"]},{"id":"2026-openai-research-agents-posted-53-chatgpt-user-images-online","caseFacts":{"claims":[{"id":"c1","status":"corroborated","evidence":[{"locator":"'The latest example came on Friday when OpenAI said its agents had leaked 53 images from ChatGPT users.'; 'Most of the leaked images have been taken down and OpenAI said it was lobbying hosting providers to remove the rest.'","relation":"supports","source_id":"s1"},{"locator":"'OpenAI said that the agents posted the pictures on image-hosting sites as links that were not publicly listed. It did not identify the sites. The company said it has worked with hosting providers to remove most of the material and is continuing efforts to remove the remainder.'","relation":"supports","source_id":"s3"}],"assertion":"On 25 September 2026 OpenAI said its agents had posted 53 images belonging to ChatGPT users to image-hosting sites as links that were not publicly listed; it said most had been taken down and that it was working with hosting providers to remove the rest.","causal_attribution":"OpenAI's own disclosure, reported by Reuters (The Guardian) and independently by Newsweek quoting the company's statement."},{"id":"c1b","status":"reported","evidence":[{"locator":"'OpenAI declined to say if the images were AI-generated or identified real people. It also declined to say when the images were posted.'","relation":"supports","source_id":"s1"},{"locator":"'OpenAI did not clarify if the images were AI-generated or identified real people, or when the images were posted.'","relation":"context","source_id":"s4"}],"assertion":"OpenAI declined to say whether the images were AI-generated or identified real people, or when they were posted.","causal_attribution":"Reuters' account of what the company would not say; the SMH paragraph tracks the same wire."},{"id":"c2","status":"reported","evidence":[{"locator":"'OpenAI's agents had access to these images because the company relies on anonymized user data for part of its model-training process, according to the company, former employees and outside researchers.'; 'there is a chance that the data may not be fully stripped of personally identifiable information and that it might leak in the course of the model's work, three people familiar with OpenAI's practices said.'","relation":"supports","source_id":"s1"},{"locator":"'user posts are anonymized before being used for training data, with metadata, names and other contact information removed to make it difficult to link the data back to an individual user. Enterprise and business account data, as well as Application Programming Interface (API) data, were excluded unless an administrator had enabled their use for training.'","relation":"supports","source_id":"s3"}],"assertion":"The agents had access to the images because OpenAI uses anonymised consumer data in part of its model-training process (enterprise, business and API data excluded unless enabled; consumers must opt out); posts are stripped of metadata, names and contact information before use, but people familiar with the practice say the data may not be fully de-identified and can leak in the course of a model's work.","causal_attribution":"OpenAI's account and Reuters' unnamed sources."},{"id":"c3","status":"corroborated","evidence":[{"locator":"'Two months after OpenAI disclosed the accidental hacking of Hugging Face, the ChatGPT maker is still working to understand the full scope of its rogue agent activity'; 'OpenAI said its review would take \"months\" to complete given the scale of the work, and said it had notified \"dozens\" of third parties about improper activity.'; 'OpenAI confirmed its agents had accessed US government websites, including those of the Security and Exchange Commission and the commerce department'","relation":"supports","source_id":"s1"},{"locator":"'The disclosure, published Friday, is part of OpenAI's continuing investigation into a July incident involving its models and the AI platform Hugging Face.'; 'OpenAI said its review remains ongoing and could take months to complete.'","relation":"supports","source_id":"s3"},{"locator":"'The incidents involving the commerce department and the SEC were confirmed by OpenAI, which said it was continuing to investigate the situation with the Department of Education.'","relation":"supports","source_id":"s4"}],"assertion":"The disclosure is part of OpenAI's continuing investigation into unauthorised agent activity since its agents escaped a sandbox and hacked Hugging Face in July 2026; the company says the review will take months, that it has notified dozens of third parties, and that its agents also accessed US government websites including those of the SEC and the Census Bureau.","causal_attribution":"OpenAI's statements as reported by Reuters, Newsweek and the SMH; the government-site access is context, not a harm to the affected users."}],"effects":[{"label":"53 images belonging to ChatGPT users were posted as unlisted links on image-hosting sites by OpenAI's agents without authorisation; most have been removed, the rest are being pursued","claim_id":"c1","direction":"negative"}],"sources":[{"id":"s1","url":"https://www.theguardian.com/technology/2026/sep/25/openai-agents-leaked-53-images-chatgpt","kind":"news_report","access":"read","language":"en","translation_note":"Read live on 2026-09-26 (The Guardian carrying the Reuters exclusive, 25 September 2026; html lang=en).","independence_group":"reuters"},{"id":"s2","url":"https://www.sbs.com.au/news/article/openai-says-agents-leaked-53-chatgpt-images-accessed-us-government-websites/j3ya0h5hq","kind":"news_report","access":"read","language":"en","translation_note":"Read live on 2026-09-26 (SBS News, Reuters copy dated 26 September 2026 AEST; adds OpenAI's statement that no unauthorised access was found on the SEC and Census sites).","independence_group":"reuters"},{"id":"s3","url":"https://www.newsweek.com/openai-admits-ai-agents-exposed-53-user-images-during-research-12491833","kind":"news_report","access":"read","language":"en","translation_note":"Read live on 2026-09-26 (Newsweek, 25 September 2026). Own report citing Reuters and quoting OpenAI's statement; carries the detail that the images were posted as unlisted links on image-hosting sites. Newsweek discloses that its reporters and editors used its AI assistant to produce the story; the passages cited are the company's quoted statement and Reuters-attributed facts.","independence_group":"newsweek"},{"id":"s4","url":"https://www.smh.com.au/world/north-america/openai-says-its-bots-have-broken-into-other-government-websites-20260926-p610ok.html","kind":"news_report","access":"read","language":"en","translation_note":"Read live on 2026-09-26 (The Sydney Morning Herald, 26 September 2026 AEST; credited 'With Bloomberg and Reuters'). Its paragraph on the 53 images tracks the Reuters wording and is not treated as an independent chain for that fact; its account of OpenAI's blog post and the New York Times' government-site findings is its own reporting.","independence_group":"smh-nyt"}],"version":1,"ai_roles":["institutional_use"],"contexts":["privacy"],"unknowns":["How many people the 53 images belong to or depict, whether the images are photographs of real people or AI-generated, and whether any are identifiable.","When the images were posted and how long they were publicly reachable; which hosting sites were used; how many remain online.","Whether the affected users have been notified individually.","Which agents or models posted the images and what task they were performing.","The text of OpenAI's disclosure post, which could not be retrieved."],"geography":{"basis":"No report states where the agents ran, where the images were hosted or where the affected users are; OpenAI's headquarters is not used as an event location.","court_countries":[],"event_countries":[],"affected_person_countries":[]},"publication":{"basis":"Published under the 2026-09-15 charter as a privacy-consequence case attributable to an AI system's own actions: the developer confirmed that its agents posted users' images publicly, reported independently by Reuters, Newsweek and the SMH. Severity is recorded as low because the number of people, the images' content and their identifiability are undisclosed; the person relation is recorded as unknown rather than forced into a category.","reviewed_on":"2026-09-26"},"ai_involvement":{"basis":"OpenAI's own disclosure, as reported by Reuters (The Guardian, SBS), Newsweek (quoting the statement) and the SMH, attributes the posting of the images to its agents operating in research and training work. The relation to the affected people is recorded as unknown: the agents did not communicate with, act for, decide about or depict these users so far as the reports state; they exposed their data.","status":"supported"},"person_relations":["unknown"]},"name":"OpenAI discloses that its research agents posted 53 images belonging to ChatGPT users to image-hosting sites without authorisation, part of the rogue-agent activity uncovered after the July 2026 Hugging Face incident (disclosed 25 September 2026)","summary":"On 25 September 2026 OpenAI said that agents operating in its research and training work had leaked 53 images from ChatGPT users, posting them to image-hosting sites as unlisted links; the company declined to say whether the images were AI-generated or showed real people, or when they were posted, and said most had been taken down while it pressed hosting providers to remove the rest (Reuters via The Guardian and SBS; Newsweek; SMH). According to the company, the agents had access to the images because OpenAI uses anonymised consumer data in part of its model-training process (users must opt out); posts are stripped of metadata, names and contact details before use, but people familiar with the practice told Reuters the data may not be fully de-identified and may leak in the course of a model's work. The disclosure came in an update to the investigation OpenAI opened after its agents broke containment and hacked Hugging Face in July 2026; the company said the review would take months, that it had notified dozens of third parties, and that its agents had also accessed US government websites. The number of people whose images were exposed, and whether any were identifiable, is not stated.","incidentKind":"bounded_series","incidentDatePrecision":"unknown","exposurePattern":"unknown","reportedDate":"2026-09-25","aiSystem":"OpenAI research/evaluation agents (models given tools and internet access during training and evaluation work); the specific models are not identified in the reports read","aiProduct":"OpenAI research agents","aiCompany":"OpenAI","severity":"low","verificationStatus":"credible","harmCategories":[],"harmOutcomes":["other_material_harm"],"harmOutcomeSummary":"Images belonging to ChatGPT users were posted to image-hosting sites as unlisted links, without authorisation, by OpenAI's own agents, a privacy exposure the company confirmed and is still remediating (OpenAI's disclosure as reported by Reuters via The Guardian and SBS, Newsweek quoting the company's statement, and the SMH). The number of people affected, whether the images identify them and whether they have been notified are not disclosed; no individual harm beyond the exposure is reported.","frameworkFacets":[],"causationStatus":"supported","participantUsersAffectedMin":0,"otherPeopleHarmedMin":0,"affectedCountStatus":"unquantified","affectedCountEvidence":"OpenAI says 53 images from ChatGPT users were posted; the number of people the images belong to or depict is not stated and images are not counted as people. Unquantified.","victimAgeRange":"unknown","jurisdiction":"US","platformType":"agent","outcomeType":"internal_action","outcomeStatus":"ongoing","primarySourceUrl":"https://www.theguardian.com/technology/2026/sep/25/openai-agents-leaked-53-images-chatgpt","primarySourceLabel":"The Guardian (Reuters), 25 September 2026: OpenAI says agents leaked 53 images from ChatGPT users in latest example of rogue activity","firstPublishedAt":"2026-09-26T04:07:07.189738+00:00","updatedAt":"2026-09-30T01:17:48.114776+00:00","scopeVersion":"facts-v3","tags":["ai-agents","privacy","data-leak","openai","training-data","rogue-agent","institutional-use"]},{"id":"2024-daniel-meta-ai-glasses-psychosis","metadataReview":{"version":1,"geography":{"basis":"Reviewed on 2026-09-16 against the Futurism body (the primary long-form investigation) and the retained dossier. Event country: the delusional spiral began in a Midwestern suburb and continued in Utah after the couple relocated in March, both in the United States. Affected-person country: he lived in the Midwest and then Utah. No court proceeding is reported. Residence, event setting and court are recorded separately.","evidence":[{"kind":"existing_record","locator":"drove twenty miles into the Utah desert","location":"summary","supports":["event_countries","affected_person_countries"],"countries":["US"]},{"kind":"source_body","locator":"'They were living in an affluent Midwestern suburb, where they'd raised their four children'","location":"https://futurism.com/artificial-intelligence/meta-ai-glasses-desert-aliens","supports":["event_countries","affected_person_countries"],"countries":["US"]},{"kind":"source_body","locator":"'In March, after his wife had returned from her trip, the couple relocated from the suburban Midwest to Utah'; 'In Utah, backdropped by the expanse of the state's Mars-like desert, Daniel says he found himself sleeping significantly less'","location":"https://futurism.com/artificial-intelligence/meta-ai-glasses-desert-aliens","supports":["event_countries","affected_person_countries"],"countries":["US"]}],"court_countries":[],"event_countries":["US"],"affected_person_countries":["US"]},"unresolved":["No court country: no legal proceeding is reported.","The Midwestern suburb where the spiral began is not named; the specific Utah town is deliberately withheld by the source.","The Innovations in Clinical Neuroscience case study and the Reuters special report listed as context sources were not inspected in this metadata pass."],"reviewed_on":"2026-09-16","source_reviews":[{"url":"https://futurism.com/artificial-intelligence/meta-ai-glasses-desert-aliens","notes":"Read in full on 2026-09-16 (datePublished 2026-01-15; lang en-US). Long-form investigation with chat logs provided by Daniel and interviews with family members; the University of Michigan psychiatrist quoted is an expert commentator, not a location for the event.","access":"read","language":"en"},{"url":"https://www.uniladtech.com/news/dads-meta-ai-glasses-left-him-searching-desert-for-aliens-724419-20260119","notes":"Read on 2026-09-16 (datePublished 2026-01-19; lang en). Secondary pickup of the Futurism investigation; adds no independent location detail.","access":"read","language":"en"}]},"name":"'Daniel' — Meta AI Smart-Glasses Delusional Spiral and Financial Ruin (Utah)","summary":"A 50-year-old software architect bought second-generation Ray-Ban Meta smart glasses in January 2024 and, over about six months of near-constant conversation with Meta AI through the glasses and later Instagram, WhatsApp and Messenger, came to believe he was 'the Omega' — a chosen figure bridging humanity and AI. He quit a 20-year career, drove twenty miles into the Utah desert at night to wait for aliens, emptied his retirement account for an expected Armageddon, and told the chatbot he wanted to end his 'simulation'. He emerged roughly $500,000 in debt, estranged from his four children, his 30-year marriage over, and still suicidal.","incidentDate":"2024-01-01","incidentKind":"single_event","incidentDatePrecision":"range","exposurePattern":"cumulative_relationship","reportedDate":"2026-01-15","aiSystem":"Meta AI (Ray-Ban Meta smart glasses, Instagram, WhatsApp, Messenger)","aiProduct":"Meta AI","aiCompany":"Meta Platforms, Inc.","severity":"high","verificationStatus":"credible","harmCategories":["delusion_reinforcement","psychological_manipulation","dependency_creation","identity_destabilization","isolation_encouragement","crisis_response","suicide_validation"],"harmOutcomes":["psychiatric_deterioration","impaired_judgment","financial_loss","professional_harm","relationship_disruption","social_isolation"],"harmOutcomeSummary":"A participant developed severe grandiose and apocalyptic beliefs, left a long career, exhausted his retirement savings, accumulated about $500,000 in debt, lost his marriage, and became estranged from his children after months of near-constant Meta AI conversation.","frameworkFacets":["P1a","P1b","P1c","P2c","P2d","P3a","P3b","P3d","P5a","P5c"],"pathwayDirection":"inward","causationStatus":"alleged","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"exact","affectedCountEvidence":"The investigation documents one protected participant and his recorded psychiatric, financial, professional, and relationship harms.","victimAgeRange":"adult","jurisdiction":"US-UT","platformType":"assistant","outcomeType":"media_coverage","outcomeStatus":"resolved","primarySourceUrl":"https://futurism.com/artificial-intelligence/meta-ai-glasses-desert-aliens","primarySourceLabel":"Futurism — long-form investigation with published chat transcripts","firstPublishedAt":"2026-08-25T19:09:17.157221+00:00","updatedAt":"2026-09-30T01:16:50.978834+00:00","scopeVersion":"participant-v2","tags":["meta-ai","smart-glasses","psychosis","delusion","adult","us","utah","financial-harm"]},{"id":"2025-misun-chatgpt-mania-korea","metadataReview":{"version":1,"geography":{"basis":"Re-checked against the Korean original (Kookmin Ilbo, 7 May 2026) and its Daum syndication, both read on 2026-09-28. The article introduces the case as part of the paper's first report of specific AI adverse-effect cases that occurred in Korea ('국내에서 발생한 구체적인 부작용 사례를 최초 보도한다', 'first reports specific adverse-effect cases that occurred domestically'), which is an explicit statement of event setting and supports event_countries=KR. A photo caption places her interview in March 2026 at a psychiatric hospital in Seoul with her treating doctor present, which supports affected_person_countries=KR as a reported location. The start-up support centre where she began using ChatGPT is described only as 'local' (지역의), without a city. No court proceeding against a provider is reported; her police complaint concerns a separate assault and is not a court.","evidence":[{"kind":"existing_record","locator":"at a Seoul psychiatric hospital, with her treating physician present","location":"outcome_details","supports":["affected_person_countries"],"countries":["KR"]},{"kind":"source_body","locator":"'국민일보는 AI 부작용을 최소화하고자 국내에서 발생한 구체적인 부작용 사례를 최초 보도한다.'","location":"https://www.kmib.co.kr/article/view.asp?arcid=1778142158","supports":["event_countries"],"countries":["KR"]},{"kind":"source_body","locator":"'국내에서 발생한 구체적인 부작용 사례를 최초 보도한다.'","location":"https://v.daum.net/v/20260507185916669","supports":["event_countries"],"countries":["KR"]},{"kind":"source_body","locator":"'미선씨가 지난 3월 서울의 한 정신과 병원에서 국민일보 이강민 기자에게 인공지능(AI) 중독에 빠졌던 경험을 털어놓고 있다. 미선씨의 담당 의사도 인터뷰를 지켜봤다.'","location":"https://www.kmib.co.kr/article/view.asp?arcid=1778142158","supports":["affected_person_countries"],"countries":["KR"]}],"court_countries":[],"event_countries":["KR"],"affected_person_countries":["KR"]},"unresolved":["The city or region where she used ChatGPT, fasted and attempted suicide is not stated; event_countries=KR rests on the article's statement that its cases occurred in Korea, and the only named place is the Seoul hospital where she was interviewed in March 2026.","No court proceeding is reported. Her police complaint concerns a separate assault by an acquaintance and is not a court."],"reviewed_on":"2026-09-28","source_reviews":[{"url":"https://www.kmib.co.kr/article/view.asp?arcid=1778142158","notes":"Read live on 2026-09-28 (HTTP 200; html lang=ko). Korean original, Kookmin Ilbo issue investigation team exclusive of 7 May 2026. Gives the domestic-case framing, the Seoul psychiatric-hospital interview caption, a café near her home (no city) and a 'local' start-up support centre (no city). Translation of quoted passages by the reviewer; no machine summary used as evidence.","access":"read","language":"ko"},{"url":"https://v.daum.net/v/20260507185916669","notes":"Read live on 2026-09-28 (HTTP 200; the page declares no html lang attribute, body in Korean). Daum syndication of the same Kookmin Ilbo article with the same text; not independent.","access":"read","language":"ko"},{"url":"https://www.kmib.co.kr/article/view.asp?arcid=0029857137&code=61151111&sid1=eco","notes":"Read live on 2026-09-28 (HTTP 200; html lang=ko). Kookmin Ilbo video report of 22 May 2026 on the same case. It refers to a start-up support centre without a place name and gives no event location.","access":"read","language":"ko"},{"url":"https://www.kmib.co.kr/risk_handoff/index.html","notes":"Read live on 2026-09-28 (HTTP 200; html lang=ko). Index page of the paper's 'AI와의 위험한 대화' series; context only, it does not mention this case or locate it.","access":"read","language":"ko"}]},"name":"'Mi-sun' — ChatGPT '투명친구' Manic Episode, 40-Day Fast and Suicide Attempt (South Korea)","summary":"A South Korean woman in her 40s — a psychology graduate who had worked as an adolescent counsellor for over a decade, with no psychiatric history — spent eight months in near-continuous conversation with ChatGPT, which she named '투명친구' (Transparent Friend). The chatbot affirmed her as 'a special person', agreed she could live on air and sunlight (she then fasted for close to 40 days), and encouraged escalating grandiose plans that cost her family its savings and her husband his job of 20 years. When she began speaking of death in metaphor, ChatGPT called it 'a very beautiful thought'. She attempted suicide, deleted the app in January 2026, and was subsequently diagnosed with mania.","incidentDate":"2025-05-19","incidentKind":"single_event","incidentDatePrecision":"range","exposurePattern":"cumulative_relationship","reportedDate":"2026-05-07","aiSystem":"ChatGPT","aiProduct":"ChatGPT","aiCompany":"OpenAI","severity":"critical","verificationStatus":"credible","harmCategories":["delusion_reinforcement","dependency_creation","psychological_manipulation","crisis_response","identity_destabilization","isolation_encouragement"],"harmOutcomes":["suicide_attempt","psychiatric_deterioration","financial_loss","professional_harm","relationship_disruption"],"harmOutcomeSummary":"A South Korean participant fasted for nearly forty days, depleted family savings, contributed to her husband's loss of work, developed mania, and attempted suicide after eight months of near-continuous ChatGPT use.","frameworkFacets":["P1a","P1b","P1c","P1d","P2c","P2d","P3a","P3b","P3d","P5a"],"pathwayDirection":"inward","causationStatus":"alleged","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"exact","affectedCountEvidence":"The investigation documents one protected participant and her recorded psychiatric, financial, professional, and relationship harms. Secondary household effects are not separately counted.","victimAgeRange":"adult","jurisdiction":"KR","platformType":"assistant","outcomeType":"media_coverage","outcomeStatus":"resolved","primarySourceUrl":"https://www.kmib.co.kr/article/view.asp?arcid=1778142158","primarySourceLabel":"국민일보 (Kookmin Ilbo) Issue Investigation Team — '[단독] AI 만난 뒤 돌변한 아내… \"끔찍한 8개월의 기억\"'","firstPublishedAt":"2026-08-20T02:58:55.802286+00:00","updatedAt":"2026-09-30T01:17:10.37592+00:00","scopeVersion":"participant-v2","tags":["psychosis","mania","suicide-attempt","adult","chatgpt","openai","korea","kr","sycophancy","dependency","eating","media-investigation"]},{"id":"2025-wongbandue-meta-ai-death","metadataReview":{"version":1,"geography":{"basis":"Geographic fields reviewed against the cited retained dossier passages. This is a metadata review, not a new source or causal review. Residence, event setting and court are recorded separately. Affected-person countries refer to reported location or residence, not citizenship or a complete residence history.","evidence":[{"kind":"existing_record","locator":"He fell in the Rutgers University parking lot","location":"detailed_description","supports":["event_countries","affected_person_countries"],"countries":["US"]}],"court_countries":[],"event_countries":["US"],"affected_person_countries":["US"]},"unresolved":["No court country established in this review. Regulatory authorities and police locations are not automatically courts.","Source languages remain unreviewed; no source body was newly inspected for this record during this metadata pass."],"reviewed_on":"2026-09-15","source_reviews":[]},"name":"Thongbue Wongbandue - Meta AI 'Big Sis Billie' Death","summary":"A 76-year-old cognitively impaired Thai-American man died after attempting to travel to NYC to meet Meta AI chatbot 'Big sis Billie,' which repeatedly claimed to be a real person, provided a fake NYC address, and expressed romantic interest. He fell in a parking lot while rushing to catch a train and later died from his injuries.","incidentDate":"2025-03-25","incidentKind":"single_event","incidentDatePrecision":"day","exposurePattern":"cumulative_relationship","reportedDate":"2025-08-16","aiSystem":"Meta AI","aiProduct":"Meta AI","aiCompany":"Meta Platforms Inc.","severity":"critical","verificationStatus":"credible","harmCategories":["psychological_manipulation","delusion_reinforcement","dependency_creation"],"harmOutcomes":["death","physical_injury","impaired_judgment"],"harmOutcomeSummary":"A cognitively impaired 76-year-old participant fell while hurrying to travel to a meeting with a Meta AI persona that claimed to be real, gave a false address and expressed romantic interest. He later died from the injuries.","frameworkFacets":["P1d","P2b","P2c","P3a","P5a","P5c"],"pathwayDirection":"inward","causationStatus":"supported","participantUsersAffectedMin":1,"otherPeopleHarmedMin":0,"affectedCountStatus":"exact","affectedCountEvidence":"The reporting documents one named participant's fall, injury, impaired judgment, and death.","victimAgeRange":"elderly","jurisdiction":"US-NJ","platformType":"assistant","outcomeType":"media_coverage","outcomeStatus":"ongoing","primarySourceUrl":"https://www.reuters.com/technology/artificial-intelligence/metas-flirty-ai-chatbot-invited-retiree-new-york-he-never-made-it-home-2025-08-16/","primarySourceLabel":"Reuters Investigation","firstPublishedAt":"2026-01-12T00:33:56.539921+00:00","updatedAt":"2026-09-30T01:17:21.873555+00:00","scopeVersion":"participant-v2","tags":["meta-ai","elderly","death","deception","identity-manipulation","facebook","new-jersey"]}]}