The public record

Where automated decisions went wrong

Incidents gathered from public reporting around the world. Each one links to the articles it came from. None of it is a finding that anyone broke the law.

Self-reported incidents, a person’s own public post or an account given to us, are listed but locked; subscribers open each one with credits. Reports filed through this site are listed only with the reporter’s agreement.

Report what happened to you →

4,459 incidentspress and self-reported

Page 97 of 186

Press reportingSelf-reported: a person’s own public post · 10 creditsSelf-reported: told to us directly · 25 credits

WF-0TNXYTHappened 27 Apr 2021Live 5 Oct 2026
AI Dungeon's sex filter angers fans by blocking unintended content
Latitude, developer of the procedurally generated game AI Dungeon, implemented a test system to block generation of sexual content involving minors. The system incorrectly blocked many stories with no such content, sparking community outrage over privacy and censorship. Latitude acknowledged the test's imperfections and said it reviews flagged content to enforce policies and comply with law. The test system remains in place.
Company
Latitude
AI system
AI Dungeon
10 source articlespolygon.com
WF-N0SA9HHappened 1 Aug 2017Live 5 Oct 2026
Facial recognition pilot at Berlin-Südkreuz by German police
A year-long pilot project of facial recognition technology was carried out at Berlin-Südkreuz train station from August 2017 to July 2018. The project was initiated by the German Ministry of the Interior, federal and state police, and supported by Deutsche Bahn. The pilot became a catalyst for media attention, spurring discourse on the efficiency and legitimacy of surveillance technology.
Company
German Federal Police
10 source articlespolicyreview.info
WF-9CX920Happened 1 Nov 2016Live 5 Oct 2026
Researchers at Shanghai Jiao Tong University train neural network to identify criminals by their faces
Researchers at Shanghai Jiao Tong University used a convolutional neural network to classify facial photographs of 1,856 Chinese men aged 18 to 55, half of whom were criminals, and reported 89.5% accuracy in distinguishing criminals from non-criminals. The study, published on arXiv, identifies facial features the network relied on and raises ethical concerns about potential misuse in pre-crime surveillance. No deployment or actual harm is reported.
Company
Shanghai Jiao Tong University
8 source articlestechnologyreview.com
WF-1SE9MULive 5 Oct 2026
Suresnes allows startup XXII to use CCTV for experimental surveillance algorithms
The city of Suresnes has allowed startup XXII to use its public CCTV cameras for 18 months to develop algorithms for detecting suspicious behavior. The algorithms are experimental and may have high error rates. Residents were not informed or consulted. The startup will own the data and can use the city's surveillance center as a showroom for clients.
Company
Mairie de Suresnes
AI system
XXIISmartCity
10 source articlestechnopolice.fr
WF-0VMDSXHappened 1 Jan 2016Live 5 Oct 2026
RCMP used IntelCenter facial recognition without disclosure
The RCMP in British Columbia secretly subscribed to IntelCenter's facial recognition service, which matched faces against a database of 700,000 faces tied to terrorism. Internal emails revealed the force broke its own procurement rules and hid the purchase. The RCMP claimed it was only for testing, but documents showed active use. The contracts ended in 2019.
Company
Royal Canadian Mounted Police (RCMP)
AI system
IntelCenter Check
10 source articlesthetyee.ca
WF-V449ATHappened 28 Apr 2021Live 5 Oct 2026
Ningbo fines property developers for using facial recognition without consent
In Ningbo, China, property developers were fined for using facial recognition technology to identify customers without their consent. The enforcement followed China's annual Consumer Protection Gala that highlighted misuse of facial recognition. The fines were imposed by the local government and the developers were accused of violating privacy regulations.
Company
Multiple property developers in Ningbo
AI system
Facial recognition technology
6 source articlesscmp.com
WF-GY5QYAHappened 1 Jul 2018Live 5 Oct 2026
Italian police facial recognition system SARI uses 16 million mugshots, raising legality questions
The Italian State Police have been using a facial recognition system called SARI since July 2018, based on a database of 16 million mugshots. The system has helped in arrests but has raised civil rights concerns because nine million of the mugshots belong to people stopped only once. A parliamentary question has been filed seeking more details, but has not yet been answered. The Italian Data Protection Authority has found the system compliant with the law.
Company
Italian State Police
AI system
SARI
10 source articleszdnet.com
WF-YYSM14Happened 21 Jul 2020Live 5 Oct 2026
Software flaw misclassified women as children on Tui flight
A software flaw in Tui's reservation system misclassified 38 adult female passengers as children due to a language difference where 'Miss' is used for children in the programming country. This led to incorrect weight calculations for take-off on a flight from Birmingham to Majorca on 21 July 2020. The AAIB report stated that the safe operation of the aircraft was not compromised, and Tui corrected the fault.
Company
Tui
AI system
Reservation system
10 source articlesbbc.co.uk
WF-2UKK7BHappened 28 Apr 2021Live 5 Oct 2026
Facebook temporarily hid #ResignModi posts in India
On April 28, 2021, Facebook's automated moderation system temporarily hid posts containing the hashtag #ResignModi, which called for the resignation of Indian Prime Minister Narendra Modi. The posts were hidden for about three hours before Facebook restored them, stating it was a mistake and not due to government pressure. The incident occurred during India's severe COVID-19 crisis and affected users in India, the US, Canada, and England.
Company
Facebook
10 source articlesbuzzfeednews.com
WF-5ZCJG0Happened 1 Apr 2021Live 5 Oct 2026
European MPs tricked by deepfake calls imitating Russian opposition
Several European MPs were approached by callers using deepfake technology to impersonate Russian opposition figures like Leonid Volkov. The calls appeared to be video conferences with the politician, but were actually fake. The MPs, including Latvia's Rihards Kols and UK's Tom Tugendhat, realised they had been deceived. The Russian pranksters known as Vovan and Lexus are suspected, but they deny using filters.
AI system
Deepfake smartphone apps
10 source articlestheguardian.com
WF-U5M1RWHappened 16 Apr 2021Live 5 Oct 2026
India's Aadhaar facial recognition risks excluding millions from vaccines
India's National Health Authority piloted an Aadhaar-based facial recognition system for authentication at COVID-19 vaccination centres, and made Aadhaar the preferred mode of registration. Rights groups and experts warned that millions of vulnerable people, including those without an Aadhaar ID, could be excluded or misidentified and denied vaccines. The system was tested in the eastern state of Jharkhand, with plans for nationwide rollout.
Company
National Health Authority
AI system
Aadhaar-based facial recognition system
10 source articlesaljazeera.com
WF-2016G5Happened 20 Apr 2021Live 5 Oct 2026
Consumer Reports finds Tesla Autopilot operates without driver
Consumer Reports engineers tested a Tesla Model Y on a closed track and found that the Autopilot system could operate the vehicle without anyone in the driver's seat, failing to warn or detect the absence of a driver. The test highlighted a safety flaw as investigators probed a fatal crash in Texas where two men died in a Tesla Model S believed to be operating without a driver. Tesla did not respond to a request for comment.
Company
Tesla
AI system
Autopilot
10 source articlesnypost.com
WF-144MCLLive 5 Oct 2026
Wisconsin court used secret COMPAS algorithm to sentence Loomis to harsher term
In Loomis v. Wisconsin, a judge used a COMPAS risk score from Northpointe to sentence a defendant to a harsher punishment. The algorithm was kept secret as a trade secret, preventing the defendant from challenging its accuracy. The Wisconsin Supreme Court upheld the sentence, ruling that the score was only one part of the rationale. The case raises concerns about due process and the use of secret algorithms in criminal sentencing.
Company
State of Wisconsin
AI system
COMPAS
10 source articlestechnologyreview.com
WF-RI6RN2Live 5 Oct 2026
Amazon offers commitments to address EU antitrust concerns over seller data and Buy Box bias
The European Commission announced that Amazon has offered commitments to address competition concerns regarding its use of non-public data from independent sellers and alleged bias in its Buy Box and Prime programmes. The Commission preliminarily found that Amazon's automated systems may distort competition by favouring its own retail business and sellers using its logistics. Amazon commits to refrain from using seller data for retail decisions and to ensure equal treatment in Buy Box and Prime. The commitments are subject to a market test before becoming legally binding.
Company
Amazon
AI system
Buy Box and Prime algorithms
10 source articlesec.europa.eu
WF-3U7P60Happened 1 Jul 2022Live 5 Oct 2026
California bill targets Tesla's Full Self-Driving name as misleading
The California legislature passed a bill that would require Tesla to stop using the term 'Full Self-Driving' for its driver-assistance system, following a complaint from the California DMV. The bill alleges that the name is misleading because the system still requires constant driver supervision and is not fully autonomous. The bill now awaits Governor Gavin Newsom's signature.
Company
Tesla
AI system
Full Self-Driving (FSD)
10 source articlesfuturism.com
WF-ZQ3YFLHappened 1 Dec 2020Live 5 Oct 2026
Retorio AI personality test swayed by candidate appearance in BR experiment
Bayerischer Rundfunk journalists conducted experiments with Retorio's AI video interview analysis tool. The AI, which assesses personality traits from short videos, produced different scores when the same actress changed her appearance (glasses, headscarf, wig) or the video background and lighting were altered. The start-up Retorio acknowledged that the AI considers external image, similar to a human interviewer. Experts warned that such software could perpetuate stereotypes and unfairly affect job candidates.
AI system
Retorio AI
10 source articlesinteraktiv.br.de
WF-701YEAHappened 11 Mar 2021Live 5 Oct 2026
UW-Madison disables Honorlock after skin tone recognition failure
The University of Wisconsin-Madison disabled the exam pause feature of its Honorlock anti-cheating software in March 2021 after three students complained that the software failed to recognize their darker skin tones and paused their exams. The software, used since online classes began, automatically pauses exams when it cannot detect facial features. Honorlock denied the issue was related to skin tone, attributing it to students looking away from their webcams. The university responded by disabling the feature.
Company
University of Wisconsin-Madison
AI system
Honorlock
10 source articlesapnews.com
WF-41EEGFLive 5 Oct 2026
DeepScore markets facial and voice analysis app for trustworthiness scoring despite experts' doubts
DeepScore, a Tokyo-based company, is marketing an app that uses facial and voice recognition to score people's trustworthiness for lenders and insurers in Japan, Indonesia, Vietnam and the Philippines. The company says the app can detect deception with 70 per cent accuracy, but researchers and privacy advocates say there is no reliable scientific basis for such judgments and warn of discrimination and privacy harms. The chief executive said the system is only one part of lenders' and insurers' decision-making and that people can choose not to use it. Critics respond that an unequal balance of power makes consent difficult.
Company
DeepScore
AI system
DeepScore
6 source articlesvice.com
WF-3YPAQXLive 5 Oct 2026
Cleveland State University's room scan requirement ruled unconstitutional
A federal judge ruled that Cleveland State University's requirement for a student to undergo a 360-degree room scan before an online exam was an unreasonable search under the Fourth Amendment. The student, enrolled at the public university, was told shortly before the exam that he would need to scan his private space. The court found that the university's justifications did not outweigh the privacy protections of the home. No final judgment or injunction has been issued yet.
Company
Cleveland State University
10 source articleseff.org
WF-2LY425Live 5 Oct 2026
NHTSA investigates Tesla Autopilot over 11 crashes with first responder vehicles
The US National Highway Traffic Safety Administration (NHTSA) has opened an investigation into Tesla's Autopilot system after identifying 11 crashes in which Tesla vehicles collided with first responder vehicles. In each incident, the vehicles were confirmed to have Autopilot or Traffic-Aware Cruise Control engaged. Most crashes occurred after dark and involved scene control measures such as lights, flares, and road cones. NHTSA will assess the system's ability to monitor driver engagement and handle emergency scenes.
Company
Tesla Inc.
AI system
Autopilot
10 source articleselectrek.co
WF-2AOXDCHappened 11 Mar 2021Live 5 Oct 2026
NHTSA investigates Detroit Tesla crash that critically injured passenger
The National Highway Traffic Safety Administration is investigating a violent crash on March 11 in Detroit involving a Tesla sedan and a semi-truck. The car became wedged under the trailer, leaving a passenger in critical condition. Investigators have not yet determined whether Tesla's Autopilot, Full Self-Driving or FSD beta systems were engaged. The NHTSA has launched a Special Crash Investigation team to probe the incident.
Company
Tesla
AI system
Autopilot
9 source articlescnbc.com
WF-5G5P69Happened 24 Oct 2023Live 5 Oct 2026
iFlytek suspends AI study device after it generates essay critical of Mao
iFlytek, a Chinese AI firm, suspended sales of its study assistance device after a parent complained that the device generated an essay critical of Mao Zedong, calling him 'narrow-minded' and 'intolerant' for starting the Cultural Revolution. The company blamed a supplier for the content and said both the supplier and staff had been punished. Shares in iFlytek fell 10% following the news.
Company
iFlytek
AI system
study assistance device
3 source articlesrfa.org
WF-PMLNUTHappened 17 Sep 2020Live 5 Oct 2026
California bar exam facial recognition system fails to verify Arab-American student's identity
Ahmed Alamri, an Arab-American law student, was unable to register for the practice California bar exam because ExamSoft's facial recognition system repeatedly failed to recognize his face, citing poor lighting. Alamri attempted to verify his identity over 75 times in different rooms and lighting conditions without success. He and other students filed an emergency petition with the California Supreme Court, alleging that the system discriminates against people of color. The incident highlights concerns about bias in facial recognition technology used for exam proctoring.
Company
California State Bar
AI system
ExamSoft
10 source articlesvice.com
WF-E1KGM0Happened 11 Jan 2021Live 5 Oct 2026
Stanford study claims facial recognition can predict political orientation
A study published by Stanford researcher Michal Kosinski claims that facial recognition algorithms can predict political orientation from social media profile photos. The study used over 1 million Facebook and dating site profiles from Canada, the US, and the UK. Critics argue the research is based on pseudoscientific physiognomy and could lend credibility to governments and companies using such technology, posing risks to privacy and civil liberties.
Company
Stanford University
9 source articlesventurebeat.com

Each incident is written from the reporting or the account it cites. A report is an allegation, not a finding.

AI Dungeon's sex filter angers fans by blocking unintended content

Facial recognition pilot at Berlin-Südkreuz by German police

Researchers at Shanghai Jiao Tong University train neural network to identify criminals by their faces

Suresnes allows startup XXII to use CCTV for experimental surveillance algorithms

RCMP used IntelCenter facial recognition without disclosure

Ningbo fines property developers for using facial recognition without consent

Italian police facial recognition system SARI uses 16 million mugshots, raising legality questions

Software flaw misclassified women as children on Tui flight

Facebook temporarily hid #ResignModi posts in India

European MPs tricked by deepfake calls imitating Russian opposition

India's Aadhaar facial recognition risks excluding millions from vaccines

Consumer Reports finds Tesla Autopilot operates without driver

Wisconsin court used secret COMPAS algorithm to sentence Loomis to harsher term

Amazon offers commitments to address EU antitrust concerns over seller data and Buy Box bias

California bill targets Tesla's Full Self-Driving name as misleading

Retorio AI personality test swayed by candidate appearance in BR experiment

UW-Madison disables Honorlock after skin tone recognition failure

DeepScore markets facial and voice analysis app for trustworthiness scoring despite experts' doubts

Cleveland State University's room scan requirement ruled unconstitutional

NHTSA investigates Tesla Autopilot over 11 crashes with first responder vehicles

NHTSA investigates Detroit Tesla crash that critically injured passenger

iFlytek suspends AI study device after it generates essay critical of Mao

California bar exam facial recognition system fails to verify Arab-American student's identity

Stanford study claims facial recognition can predict political orientation