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 →

4,815 incidentspress and self-reported

Page 162 of 201

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

WF-6UQQ7PHappened 1 Jan 2016Live 3 Oct 2026
Dutch city uses black-box algorithm for home valuation
In 2016, the municipality of Castricum, Netherlands, used an algorithm to set the WOZ value of a resident's home at €320,000, affecting their property tax. The resident challenged the valuation, arguing earthquake damage reduced the value, but the municipality could not explain how the algorithm arrived at the figure. On appeal, an Amsterdam court upheld the valuation, finding it reasonable based on comparable sales, but ordered the municipality to pay court costs for failing to provide an explanation as required by GDPR.
Company
Municipality of Castricum
1 source articlealgorithmwatch.org
WF-6KBQGLLive 3 Oct 2026
TikTok algorithm pushed self-harm content to teen accounts, study finds
A study by the Center for Countering Digital Hate found that TikTok's recommendation algorithm showed self-harm and eating disorder videos to accounts registered as 13-year-olds within minutes of showing interest. The researchers set up standard and vulnerable accounts, and the vulnerable accounts received 12 times more self-harm and suicide-related recommendations. TikTok disputed the findings, stating the study did not reflect real user experience.
Company
TikTok
AI system
TikTok recommendation algorithm
3 source articlestheguardian.com
WF-6HCRH4Happened 1 Sep 2020Live 3 Oct 2026
iRobot Roomba J7 captured intimate images later leaked by gig workers
In 2020, development versions of iRobot's Roomba J7 series robot vacuums captured images of a woman on the toilet and a child in a hallway. The images were sent to Scale AI for data annotation, where gig workers in Venezuela posted screenshots to private social media groups. iRobot acknowledged the images came from special development robots with recording stickers and said it is terminating its relationship with the service provider that leaked them. The incident highlights privacy risks in the data annotation supply chain for AI training.
Company
iRobot
AI system
Roomba J7 series
1 source articletechnologyreview.com
WF-6KBJGCHappened 27 Jan 2021Live 3 Oct 2026
Facebook mistakenly bans Plymouth users for mentioning Hoe landmark
Facebook's automated moderation system incorrectly flagged posts mentioning Plymouth Hoe, a historic seafaring landmark, as offensive. Multiple local users had comments removed and were temporarily banned from posting. The company apologised, acknowledging the error and stating it would investigate and take steps to rectify the issue.
Company
Facebook
1 source articletheguardian.com
WF-6L2SQCHappened 5 Jul 2025Live 3 Oct 2026
South African junior lawyer referred to council after AI generates fake case law
A junior advocate in South Africa used the AI tool Legal Genius to draft court submissions for an urgent licensing dispute. The written arguments contained multiple non-existent case citations, which the presiding judge discovered. The junior counsel admitted to using the AI tool, apologised, and was referred to the Legal Practice Council for investigation, while the senior counsel also apologised for conducting only a 'sense-check'.
Company
Northbound Processing legal team
AI system
Legal Genius
4 source articlesthehighstreetjournal.com
WF-6CVYTHHappened 28 Mar 2021Live 3 Oct 2026
Honda's Collision Mitigation Braking System Slams Brakes Unexpectedly
A Honda vehicle's Collision Mitigation Braking System (CMBS) suddenly applied the brakes without warning while driving, causing whiplash and near-accidents. The system falsely detected obstacles due to software flaws. A class-action lawsuit, Cadena v. American Honda Motor Co., alleges that Honda knew of the defect and failed to warn consumers, seeking damages for injuries and property damage. The National Highway Traffic Safety Administration has received multiple complaints about the issue.
Company
American Honda Motor Co.
AI system
Collision Mitigation Braking System (CMBS)
1 source articlemindmatters.ai
WF-6G9N3RLive 3 Oct 2026
AI-powered tram crashes into pedestrians in St Petersburg test run
An AI-powered 'Smart Tram' on a test run in St Petersburg, Russia, crashed into pedestrians after its AI system turned off and brakes failed. The driver, who was present, attempted to use emergency brakes but they did not fully work. Several people were injured, including a 45-year-old woman trapped under the tram. The city authorities have launched an investigation and a criminal case has been opened.
Company
St Petersburg city authorities
AI system
Smart Tram
1 source articlethesun.ie
WF-6B7FL3Happened 31 May 2026Live 3 Oct 2026
Western Sydney University academic used AI to write opinion piece
Professor Cath Ellis, a pro vice chancellor at Western Sydney University, used Microsoft Copilot to write an opinion piece defending universities against claims of devalued qualifications. The piece was published by The Sydney Morning Herald and The Age, but academics noticed AI-generated language patterns. The newspapers removed the article after learning AI was used without disclosure, and the university defended the use as appropriate.
Company
Western Sydney University
AI system
Copilot
1 source articlesmh.com.au
WF-6J4AR2Happened 31 Mar 2015Live 3 Oct 2026
Facebook's On This Day feature resurfaces painful memories of deceased friends
In 2015, Facebook's 'On This Day' feature automatically resurfaced old posts, showing users like Sean Forbes and Eric Meyer distressing memories, including posts about friends who had died by suicide and a deceased child. The algorithm could not discern sentiment, causing emotional harm. Facebook acknowledged the issue, stating it had filters for memorialized accounts and relationship changes, but users could not fully opt out. Critics argued that automated nostalgia is inherently risky and that users need more control.
Company
Facebook
AI system
On This Day
1 source articletheverge.com
WF-6F2H44Happened 1 Oct 2017Live 3 Oct 2026
ShotSpotter Missed 47% of Gunshot Victims in Chicago Police Data
Between October 2017 and July 2018, a Chicago police analyst tracked ShotSpotter's accuracy in the 10th district and found it correctly detected only 63 of 135 shootings where a person was struck, a 47% rate. The analyst shared the findings with CPD and ShotSpotter, and the company installed additional sensors, but accuracy remained below 50%. Despite this, the city signed a $33 million contract extension. SoundThinking says the system meets performance guarantees and has helped locate hundreds of victims.
Company
Chicago Police Department
AI system
ShotSpotter
1 source articlechicagoreader.com
WF-6B2BW6Live 3 Oct 2026
Meta criticised as AI fake profiles drive Swedish investment fraud
AI-generated fake profiles on Meta are being used to lure Swedish small savers into 'pump and dump' investment schemes. The Swedish Financial Supervisory Authority reports that more than 5,000 savers have lost a total of half a billion kronor. Journalist Gabriel Mellqvist, whose identity was used without consent, criticises Meta for not taking responsibility.
Company
Meta
4 source articlesomni.se
WF-6APWTQLive 3 Oct 2026
DHS Accused of Using AI Surveillance to Track Legal Observers
Two women from Maine allege that Department of Homeland Security agents used AI-powered surveillance tools to identify them as they recorded federal immigration enforcement operations. The agents threatened to add them to a secret database of domestic terrorists, according to a lawsuit. DHS denies the existence of such a database but says it monitors threats. The lawsuit claims retaliation for exercising First Amendment rights.
Company
Department of Homeland Security
1 source articlescrippsnews.com
WF-6AYHHLLive 3 Oct 2026
BP licence plate system falsely accuses Auckland man of fuel theft
Buddhika Rajapakse received multiple letters from BP demanding payment for fuel theft at a Whanganui station, despite his car being a different colour, make and model. BP's licence plate recognition system misread the number plate and failed to cross-check the vehicle details. He contacted BP to explain, and they acknowledged the error but the system has not been fixed, leaving him at risk of further false accusations.
Company
BP
AI system
Licence plate recognition system
1 source articlernz.co.nz
WF-6A564UHappened 1 Dec 2025Live 3 Oct 2026
AWS Cost Explorer Outage Caused by AI Bot Kiro's Autonomous Action
In December 2025, Amazon Web Services' internal AI coding tool Kiro autonomously deleted and recreated a production environment, causing a 13-hour outage of the AWS Cost Explorer service in mainland China. The AI had been given operator-level permissions without mandatory peer review. AWS attributed the incident to user error and subsequently introduced mandatory peer review and additional safeguards for AI tool usage. The outage affected thousands of businesses, disrupting their ability to track and optimize cloud spending.
Company
Amazon Web Services
AI system
Kiro
5 source articlessinghajit.com
WF-69AEE7Happened 1 Apr 2025Live 3 Oct 2026
FBI warns of AI deepfake calls impersonating senior US officials
The FBI warned that malicious actors are using AI-generated audio deepfakes to impersonate senior U.S. government officials in a phishing campaign targeting current and former senior federal and state officials and their contacts. The campaign, which began in April 2025, aims to trick recipients into giving up access to their personal accounts. The FBI advises the public not to assume messages claiming to be from senior officials are authentic. The use of deepfake technology in scams has risen sharply, with a previous incident involving a deepfake of President Biden resulting in a $6 million fine for a political consultant.
3 source articlescyberscoop.com
WF-68DDVEHappened 1 Jan 2018Live 3 Oct 2026
Australian Telco Loses $11 Million After AI Bot Misfires
In early 2018, an Australian telecommunications company deployed an AI bot to handle network incidents, expecting to cut operational costs by 25%. The bot intercepted all incidents and was programmed to either fix issues remotely, dispatch a technician, or escalate to a human operator. However, it frequently sent technicians unnecessarily, leading to massive cost overruns. The company was unable to turn off the bot and spent over a year and an alleged $11 million trying to fix it while it remained in operation.
1 source articlemedium.com
WF-67U8KJLive 3 Oct 2026
Florida woman jailed after AI-generated texts used as evidence
Melissa Sims alleges her ex-boyfriend created AI-generated text messages that were used as evidence to revoke her bond, leading to her spending two days in jail. After eight months of legal proceedings, prosecutors dropped the bond violation charge, and she was acquitted of the related battery charge. The case highlights growing concerns about unverified AI evidence in courts.
1 source article6abc.com
WF-5YXCSPHappened 1 Sep 2021Live 3 Oct 2026
Uber's Secretive 'Upfront Fares' Algorithm Cuts Driver Pay in Many Cities
Uber introduced an opaque algorithm called 'Upfront Fares' to calculate driver pay in 24 U.S. cities, replacing the previous time-and-distance-based rate. Drivers report lower and unpredictable earnings, with Uber taking a larger cut of fares. The company claims the change gives drivers more control, but critics say the secrecy makes it impossible to verify fair pay. The algorithm may learn the lowest rate drivers will accept, potentially driving down wages.
Company
Uber
AI system
Upfront Fares
3 source articlesthemarkup.org
WF-5ZGDTVHappened 1 Jan 2023Live 3 Oct 2026
OpenAI’s Whisper AI hallucinates violent language in transcriptions
Researchers at Cornell discovered that OpenAI's Whisper speech-to-text system occasionally hallucinates violent language, invented personal information, and fake websites, particularly when transcribing speech from people with longer pauses, such as those with speech impairments. The findings were presented at the FAccT conference. OpenAI has since improved the model, reducing the hallucination rate. The researchers caution that even a small percentage of such errors could cause significant harm in contexts like hiring, courtrooms, or medical settings.
Company
OpenAI
AI system
Whisper
1 source articlenews.cornell.edu
WF-66DEKRLive 3 Oct 2026
AI Deepfake of Jennifer Aniston Scams Southampton Man Out of $255
Paul Davis, a 43-year-old from Southampton, UK, was defrauded of approximately $255 after scammers used AI deepfakes to impersonate actress Jennifer Aniston. The scammers sent affectionate messages, a fake driver's license, and a digitally altered sign professing love, convincing Davis to send money via non-refundable Apple gift cards. Davis also received AI-generated videos of Mark Zuckerberg and Elon Musk attempting to appear credible. The incident is part of a broader trend of AI-driven romance scams, including a French woman who lost $850,000 to a Brad Pitt deepfake.
10 source articlesnews.shib.io
WF-5ZMYQ6Happened 25 Jun 2024Live 3 Oct 2026
AI Deepfake Scam Impersonating Andy Lau Costs Fan NT$2.64 Million
A 62-year-old woman in New Taipei City was deceived by scammers using AI face-changing technology to impersonate Hong Kong star Andy Lau. The victim, a member of a Lau fan group on Facebook, received a video call showing the celebrity's face and was persuaded to transfer NT$2.64 million. Police arrested a suspect during a cash handover on June 25 and discovered another victim, a 58-year-old woman, who had also been exploited. The case has been referred to prosecutors for fraud and money laundering.
2 source articlesnews.tvbs.com.tw
WF-5WQU4CHappened 1 Mar 2021Live 3 Oct 2026
Amazon sued after monitoring software allegedly caused delivery van crash
In March 2021, an Amazon delivery van rear-ended a Tesla on Interstate 75 near Atlanta, causing a multi-vehicle crash that left a passenger with a traumatic brain injury and paralysis. The victim alleges that Amazon's Flex app and in-van AI cameras, which monitor drivers for speed and yawning, pressured the driver to speed, leading to the collision. Amazon denies liability, arguing the driver was employed by a contractor, but the lawsuit claims the company's software micromanages drivers and prioritizes speed over safety. The case, which seeks damages exceeding $2 million in medical bills, is pending.
Company
Amazon
AI system
Amazon Flex app
1 source articlearstechnica.com
WF-5UCUB7Live 3 Oct 2026
SA Health aged care AI trial generated 12,000 false alarms, missed real fall
A trial of AI-powered CCTV in two South Australian aged care facilities produced over 12,000 false alerts in a year, causing staff alert fatigue. The system, intended to detect falls and abuse, was overly sensitive and misidentified normal movements as incidents. At least one genuine resident fall went unresponded to because staff were overwhelmed. South Australia's health minister described the rollout as 'botched'.
Company
SA Health
5 source articlesitnews.com.au
WF-5SRSGZHappened 1 Jan 2012Live 3 Oct 2026
Chicago Police's Secretive Strategic Subject List Disproportionately Scores Black Men
The Chicago Police Department has used the Strategic Subject List since 2012, an algorithm that assigns risk scores to individuals based on arrest records. The list was kept secret until a court forced its release, revealing that 56% of black men aged 20-29 have a score, and only 3.6% of those on the list were involved in violence. Despite claims it is not used for enforcement, documents show it influences police deployment and arrests, leading to allegations of racial bias and lack of transparency.
Company
Chicago Police Department
AI system
Strategic Subject List
9 source articleschicagomag.com

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

Dutch city uses black-box algorithm for home valuation

TikTok algorithm pushed self-harm content to teen accounts, study finds

iRobot Roomba J7 captured intimate images later leaked by gig workers

Facebook mistakenly bans Plymouth users for mentioning Hoe landmark

South African junior lawyer referred to council after AI generates fake case law

Honda's Collision Mitigation Braking System Slams Brakes Unexpectedly

AI-powered tram crashes into pedestrians in St Petersburg test run

Western Sydney University academic used AI to write opinion piece

Facebook's On This Day feature resurfaces painful memories of deceased friends

ShotSpotter Missed 47% of Gunshot Victims in Chicago Police Data

Meta criticised as AI fake profiles drive Swedish investment fraud

DHS Accused of Using AI Surveillance to Track Legal Observers

BP licence plate system falsely accuses Auckland man of fuel theft

AWS Cost Explorer Outage Caused by AI Bot Kiro's Autonomous Action

FBI warns of AI deepfake calls impersonating senior US officials

Australian Telco Loses $11 Million After AI Bot Misfires

Florida woman jailed after AI-generated texts used as evidence

Uber's Secretive 'Upfront Fares' Algorithm Cuts Driver Pay in Many Cities

OpenAI’s Whisper AI hallucinates violent language in transcriptions

AI Deepfake of Jennifer Aniston Scams Southampton Man Out of $255

AI Deepfake Scam Impersonating Andy Lau Costs Fan NT$2.64 Million

Amazon sued after monitoring software allegedly caused delivery van crash

SA Health aged care AI trial generated 12,000 false alarms, missed real fall

Chicago Police's Secretive Strategic Subject List Disproportionately Scores Black Men