Skip to main content

NOPE · AI and people

AI incidents

Reports of AI-related harm and adverse experiences affecting people’s safety, wellbeing, rights and livelihoods. Explore what happened and the evidence available.

NOPE’s core concern is when AI communicates with a person, acts on their behalf, or depicts or impersonates them. The tracker also records consequential decisions, claims and privacy harms involving AI. Each case needs a described connection between AI use and the harm, including private information recorded into or disclosed to an AI service. Each account is reviewed for publication; claims may remain uncorroborated or disputed. How we review and count cases

In this selection

Published cases
10
Countries with reported events
7
Located 10 of 10 cases · 0 unknown
Languages in checked sources
4
Recorded for 10 of 10 cases

1 case has no reviewed AI-to-person relation yet: 0 not yet reviewed and 1 reviewed as unknown. Show these cases

These figures describe the cases collected by NOPE. Coverage varies with discovery, reporting and available evidence. They do not estimate how often AI-related harm occurs.

Response counts currently use each case’s principal recorded outcome. Further proceedings may be described in its account.

Cases in this selection, counted once in their first known event year. A series may continue beyond that year. Reporting and collection dates are excluded. NOPE has searched recent events more thoroughly than earlier years, so bar heights also reflect collection effort.

Reported severity MediumLow
More filters: AI relation, use, setting, sources and responses
Clear filters

10 of 1102 published cases

Jul 2025IndiaPoshan Tracker face recognition

India: pregnant women and new mothers report being cut off from take-home rations when the Poshan Tracker app's face match failed

From 1 July 2025 India's Ministry of Women and Child Development made a facial recognition check in the government's Poshan Tracker app mandatory before pregnant women and nursing mothers receive take-home rations under the Integrated Child Development Services (ICDS) scheme; the app compares the woman's live face with her Aadhaar identity photograph. In July 2025 Decode reported from Delhi that a woman eight months pregnant returned to her Anganwadi centre for a third day trying to get the app to recognise her, and a worker said she gave food to a new mother the app would not recognise and recorded it by hand. In August 2025 the Karnataka Anganwadi workers' union told The South First that the faces of pregnant and nursing mothers change so they no longer match their registration images, and that some mothers leave without their rations. In March 2026 a Decode investigation supported by the Pulitzer Center reported that a mother in Jharkhand was cut off from supplementary nutrition ten weeks after giving birth because the software could not match her face to a photograph taken eight years earlier, that a pregnant woman in Bihar who was anaemic went four months without rations because the system could not match her to an older Aadhaar photograph, that one Bihar centre had removed three pregnant women and five nursing mothers whose faces no longer matched, and that a Jharkhand worker was told to remove at least four children whose eligibility depended on their mothers' failed face matches. Decode also reports failures earlier in the process, when women cannot receive a one-time password on an Aadhaar-linked phone. Government figures cited by Decode show 52.7% of eligible beneficiaries had received rations through facial recognition by 31 December 2025. A researcher who analysed the app for Decode found Google's ML Kit face-detection library in it; Google said ML Kit cannot recognise individuals, and the government has not disclosed who built the matching system. After workers' protests, a February 2026 app update let workers upload other identity documents for pregnant and nursing women, subject to a supervisor's approval. Decode reports the Bihar woman later gave birth to a healthy baby, while her name remained missing from the system.

Contextual tracker case Medium reported severity Internal Action

AI involvement reported · Causal attribution alleged · 4 sources · Added 02/11/2026

Oct 2023NetherlandsMONOcam (reported)

Dutch driver says police AI phone camera flagged him for holding a phone while he scratched his head, and an officer approved a 380-euro fine

Dutch police use mobile smart cameras that apply artificial intelligence to photos of passing drivers to detect whether they are holding a mobile phone; a police officer then reviews each flagged photo before a fine is issued. A driver says that in October 2023, on the A2 motorway near Weert, such a camera photographed him with his right hand at his head and that he received a fine of 380 euros plus 9 euros in administration costs for holding a mobile device while driving. He says the flash photo, which he requested from the Central Judicial Collection Agency (CJIB), shows his empty hand near his face: he was scratching his head, his lease car allows hands-free calling, and he is left-handed. In a February 2024 blog post he writes that the camera was a MONOcam, that its algorithm judged that he was holding a device and that the reviewing police employee accepted that judgement, so that both the algorithm and a person erred; RTL Nieuws reported that the photo was probably taken by a MONOcam. Omroep Brabant described the fine as a mistake by the AI camera. The driver said the reviewing officer was probably biased by the algorithm's output and that a dark shadow in his hand could mislead a quick look. He lodged an objection with the Public Prosecution Service; it had not been decided when the reports appeared in February 2024, and a CJIB spokesperson, who would not comment on individual cases, suggested to RTL that he was unlikely to have to pay. No later outcome has been reported. He also told Omroep Brabant that he considers the cameras valuable for road safety, provided doubtful cases are not fined.

Contextual tracker case Low reported severity Media Coverage

AI involvement reported · Causal attribution alleged · 3 sources · Added 01/11/2026

2024United StatesHayden AI bus-lane cameras (reported)

New York MTA says its AI bus-lane cameras wrongly ticketed drivers parked in legal spots and voided about 3,800 mistaken violations

New York's Metropolitan Transportation Authority (MTA) uses cameras powered by artificial intelligence, mounted on bus windshields, to catch vehicles blocking bus lanes; the system analyses images to decide whether it sees a vehicle stopped where it should not be, and the city's Department of Transportation (DOT) says a person reviews every machine-generated violation before a ticket is issued. In November 2024 NBC New York reported that, during the 2024 expansion to new routes, the cameras on the M79 and Bx35 routes had mistakenly ticketed about 3,800 vehicles. According to the MTA, more than 870 of those violations went to cars, trucks and SUVs parked in legal alternate-side spots, because the cameras had not been programmed to skip those zones; most of the rest were issued during what should have been a warning period. One driver told NBC New York and NPR that tickets for parking he knew to be legal kept arriving, about a dozen in all and three in one day, costing hundreds of dollars; he said he tried to fight them in court, which was not working well, and received apology letters only after his story was broadcast. A second driver's summons carried a photo showing him in a legal alternate-side spot. The MTA said all the mistaken violations were reversed, payments were being refunded and the software was corrected. An unnamed source told NBC the cameras were made by Hayden AI; the MTA blamed about 800 of the tickets on a Hayden AI programming error, and a Hayden AI officer told NPR the problems had nothing to do with the AI or the technology and were a configuration issue. DOT declined to say how its human reviewers missed the erroneous violations.

Contextual tracker case Low reported severity Internal Action

AI involvement reported · Causal attribution supported · 2 sources · Added 31/10/2026

Jan 2023MexicoCBP One app

Advocates say CBP One asylum app's facial recognition rejected photos of Black asylum seekers at the US-Mexico border, blocking appointments

In January 2023 US Customs and Border Protection (CBP) made its CBP One mobile app the only way for migrants at the US-Mexico border to apply for asylum appointments and exemption from Title 42 restrictions; applicants had to upload a photo of themselves through the app. In February 2023 the Guardian and the Border Chronicle, both reporting by Melissa del Bosque, quoted three non-profits working with Haitian and African asylum seekers in Tijuana, Reynosa and Matamoros who said the app's facial recognition was failing to register many people with darker skin, so they could not upload their photos and obtain appointments. The executive director of Al Otro Lado said Haitians at a workshop her organisation held in Tijuana the day after the app went live got error after error message. The Sidewalk School said hardly anyone among about 4,000 Black asylum seekers in Reynosa and 1,000 Haitians in Matamoros was getting an appointment, and that shining construction lights on faces let adults get past the photo step but not children under six, which kept families from applying. A Haitian Bridge Alliance attorney said the app had helped lighter-skinned applicants but not Haitians. The advocates also described other barriers, such as outdated phones and poor internet access. CBP did not answer the Guardian's questions before publication and told the Border Chronicle no one was available to respond. No technical analysis of the failures is reported.

Contextual tracker case Low reported severity Media Coverage

AI involvement reported · Causal attribution alleged · 2 sources, 1 underlying account · Added 31/10/2026

Mar 2021United StatesID.me Face Match

Nevada unemployment claimant says ID.me face check rejected him three times and locked him out, leaving him three weeks without benefits

ID.me, an identity-verification company used by at least 21 US states to check people claiming unemployment benefits, asks claimants to pass a facial recognition step that compares their face with a photo of the same person, such as a driving licence. In June 2021 Motherboard (VICE) reported that a gig economy worker in Las Vegas was suddenly cut off from his unemployment benefits in late March after ID.me failed to identify him. He said that when he held his phone in the instructed position the system rejected him three times without giving a reason and then locked him out. He said he tried ID.me's chat support many times and the state of Nevada sent him back to ID.me; after several weeks he posted a public criticism on Twitter, and the company contacted him and, after several more days, verified his identity. He said he went three weeks without his benefit and could not pay bills. ID.me's chief executive told Motherboard that its Face Match algorithms operate at about 99.9% efficacy, that failures are a problem with how people use the system rather than the technology, and that the company was unaware of eligible people unable to verify their identity. Motherboard said other claimants in several states described similar lockouts; those accounts are not counted here.

Contextual tracker case Low reported severity Media Coverage

AI involvement reported · Causal attribution alleged · 2 sources · Added 30/10/2026

Apr 2019United KingdomHM Passport Office photo checker

UK passport photo checker told Black applicants their closed mouths looked open, they say; a BBC test found darker-skinned women flagged more often

The UK government's online passport application uses an automated face-detection check that tells applicants when it thinks an uploaded photo breaks the rules, which include a neutral expression and a closed mouth. In September 2019 PA reported that a young Black man renewing his passport online said the checker mistook his lips for an open mouth, and that to submit the photo anyway he had to write in a comment box that his mouth was closed and he just had big lips. PA also reported that a Black educational technologist said the system told her that her eyes looked closed and that it could not find the outline of her head; she told the BBC she uploaded 10 photographs over a week and each was rated poor, and PA reported that the Passport Office tweeted that it was sorry the service had not worked as it should. In October 2020 a BBC investigation reported that a Black student said she was wrongly told her mouth looked open on each of five photos and got one accepted only by writing a note challenging the verdict. The BBC fed 1,130 passport-style photos of politicians into the checker and found that photos of dark-skinned women were graded poor quality 22% of the time against 14% for light-skinned women; it reported that documents released under a 2019 freedom of information request showed the Home Office knew of the problem but judged overall performance good enough to launch. The Home Office said applicants can override the check, that the checker is a customer aid, and that its systems are improving.

Contextual tracker case Low reported severity Media Coverage

AI involvement reported · Causal attribution supported · 2 sources · Added 29/10/2026

25 May 2025IsraelUnidentified AI tool

Israel: Supreme Court finds Ramat Gan refused a special-education pupil school transport in a letter resting on uncontrolled AI use and invented rulings

In May 2025 the Ramat Gan municipality in Israel refused a request that a special-education pupil be given school transport from his father's home, citing an Education Ministry director-general circular and quotations from recent rulings. The father's lawyer could not find the circular, and the Education Ministry said it did not know it. After the father petitioned the Tel Aviv District Court, the municipality agreed in July 2025 to restore the transport but again filed invented quotations, and the district court refused costs. On appeal, the Supreme Court (Deputy President Noam Sohlberg, ruling published 22 March 2026) held that the documents left no doubt that the municipality had used artificial intelligence and that its decision, at least as presented to the pupil, clearly rested on uncontrolled AI use. It ordered the municipality to pay 30,000 shekels in costs.

Contextual tracker case Medium reported severity Involving minors

AI involvement supported · Causal attribution established · 3 sources · Added 27/10/2026

2021IndiaSamagra Vedika

Al Jazeera investigation says Telangana's Samagra Vedika system wrongly matched poor families to car owners, and officials denied them food subsidy cards

An Al Jazeera investigation published on 24 January 2024, produced with The Reporters' Collective and the Pulitzer Center's AI Accountability Network, reports that Samagra Vedika, a system the Telangana government uses to consolidate residents' data and check welfare eligibility, wrongly tagged poor families in Hyderabad as owning cars, and that officials relied on those tags to deny them food security cards (subsidised rice). Al Jazeera says the system uses machine learning and entity resolution to match people across databases, and that since 2016 officials have been mandated to check whether the system approved an applicant before deciding. In one case it reports that the system mistook a widow's late husband, a rickshaw puller, for a car owner with a similar name; her renewal application was rejected in 2021 for possessing a four-wheeler. After she traced the vehicle's real owner, officials agreed there had been an algorithmic error but said her family income exceeded the limit, which Al Jazeera says was not the case; she had been without the subsidised rations for more than seven years. A second family, whose husband cannot drive, was rejected in 2021 for possessing a four-wheeler; in November 2023 the High Court of Telangana said the petitioner is eligible whenever new cards are issued. The government says the technology is highly precise and that wrong vehicle or house tags occur in under five per cent of cases and have mostly been corrected.

Contextual tracker case Medium reported severity Media Coverage

AI involvement supported · Causal attribution supported · 1 source · Added 24/10/2026

20 Aug 2026SingaporeGemini (reported)

Singapore: an AI-generated photo of a crocodile at Pandan Reservoir led PUB to suspend water activities and search the reservoir in August 2026; a contractor's worker who allegedly made it was charged with transmitting a false message and with obstructing justice for allegedly deleting the image and the Gemini app history

On 20 August 2026 Singapore's national water agency PUB received a report of a crocodile sighting at Pandan Reservoir, with a photo. PUB said it tried to verify the image with several AI checker tools without a definite result, so it suspended water activities, told the Singapore Rowing Association and Singapore Canoe Federation to stop all activities there, put up warning signs and, with the National Parks Board, ran search-and-capture operations. On 21 August PUB found the photo was fake, stopped searching and filed a police report; water activities resumed on 22 August. Police said on 28 September that preliminary investigations found a man had used artificial intelligence to generate the image, sent it to a colleague, and then deleted the image and the related application to conceal his actions. On 29 September the 30-year-old, a worker employed by PUB's reservoir maintenance contractor, was charged with transmitting a false message and with obstructing the course of justice; the charge sheets, as reported by The Straits Times, say he deleted the image and the Gemini AI application history log from his phone on 21 August. He told the court he intended to plead guilty, and the case returns to court in November.

AI relation unknown Low reported severity Criminal Charges

AI involvement reported · Causal attribution alleged · 5 sources, 2 underlying accounts · Added 02/10/2026

15 Sept 2026IndiaUnidentified traffic-violation detection camera

Bengaluru: an AI traffic camera on the Outer Ring Road reportedly classified a guitar bag on a lone scooter rider's back as a helmetless pillion passenger and generated a ₹500 e-challan; a senior traffic officer said the system appears to have made the classification, and police offered rectification

On 15 September 2026 a man was riding his wife's electric scooter alone in Bengaluru, from Hebbal towards Tin Factory, wearing a helmet and carrying a guitar on his back. An AI-enabled Bengaluru Traffic Police camera on the Outer Ring Road photographed him, and the next day an e-challan of ₹500 was generated for a pillion rider not wearing protective headgear; the photograph attached to it showed no passenger, only the guitar bag. The rider posted the challan on X and asked the traffic police to revoke it. A senior traffic police officer told Deccan Herald that the bag's position apparently led the AI-based system to classify it as a person sitting behind the rider, after which the system appears to have detected a helmet violation and generated the challan. Bengaluru Traffic Police replied that the violation could be rectified and gave an email address and phone numbers, and the joint commissioner of police (traffic) told the Times of India that the violation would be revoked if the complaint was found true. Whether it was revoked is not reported.

Contextual tracker case Low reported severity Media Coverage

AI involvement supported · Causal attribution supported · 4 sources, 3 underlying accounts · Added 02/10/2026

Cases may have several effects and sources. Mixed accounts qualify when they include a reported harm or adverse experience. People are counted within individual cases where sources support a number; we do not publish a collection-wide total of distinct people.

A source’s existence, the experience it reports and AI’s causal role are separate questions. A lawsuit records allegations unless a subsequent finding establishes them.

Methodology and corrections · Subscribe via RSS · Suggest a case or correction · Find support

Last dataset update: 18/11/2026. Dataset available under CC BY 4.0.