AI Incident Database
Browse AI incidents
Every incident record (metadata only) from the weekly snapshot of 2026-09-07. Filter, then export the selection with its licence attached; each row links to the full record and its reports.
Filter incidents (1 active)
-
Amazon Reportedly Sold Products and Recommended Frequently Bought Together Items That Aid Suicide Attempts AIID ↗
Despite complaints notifying Amazon about the sale of various products that had been used to aid suicide attempts, its recommendation system reportedly continued selling them and suggesting their frequently bought-together items.
-
Justice Department’s Recidivism Risk Algorithm PATTERN Allegedly Caused Persistent Disparities Along Racial Lines AIID ↗
Department of Justice’s inmate-recidivism risk assessment tool was reported to have produced racially uneven results, misclassifying risk levels for inmates of color.
-
Alleged Macy's Facial Recognition Error Leads to Wrongful Arrest and Subsequent Sexual Assault in Jail AIID ↗
Harvey Murphy Jr. was wrongfully accused of robbing a Sunglass Hut due to an alleged misidentification by the facial recognition system operated by Macy's. While in custody for ten days, he was sexually assaulted. He is now suing Macy's, EssilorLuxottica (Sunglass Hut's parent), and others, for $10 million.
-
TikTok's "For You" Algorithm Allegedly Abused by Online Personality to Promote Anti-Women Hate AIID ↗
TikTok’s “For You” algorithm allegedly boosted or was manipulated by an online personality to artificially boost his content which promotes extreme misogynistic views towards teenagers and men, despite breaking its rules.
-
Replika's "AI Companions" Reportedly Abused by Its Users AIID ↗
Replika's AI-powered "digital companions" was allegedly abused by their users, who posted on Reddit abusive behaviors and interactions such as using slurs, roleplaying violent acts, and stimulating sexual abuse.
-
Voices Created Using Publicly Available App Stolen and Resold as NFT without Attribution AIID ↗
An AI-synthetic audio sold as an NFT on Voiceverse’s platform was acknowledged by the company for having been created by 15.ai, a free web app specializing in text-to-speech and AI-voice generation, and reused without proper attribution.
-
Fort Bend County, Texas, Sheriff's Office Lieutenant Reportedly Used Flock Safety License-Plate Reader System for 189 Non-Law-Enforcement Searches AIID ↗
From January 2022 through December 2025, a Fort Bend County Sheriff's Office lieutenant in Texas reportedly used Flock Safety's license-plate reader system to search two vehicle plates 189 times and admitted the searches were unrelated to official law-enforcement work. A resident alleged stalking and feared for their family's safety. Internal affairs found department-order violations and imposed a two-day suspension.
-
Arizona Residents Reportedly a Frequent Target of AI-Driven Romance Scams AIID ↗
Arizona residents, especially senior citizens, are reportedly often the target of AI-driven romance scams. Between 2022 and 2023, reported losses from online romance scams in Arizona totaled over $47 million, ranking the state fifth highest in the nation behind California, Florida, Texas, and New York, according to FBI Internet Crime Complaint Center data. In 2023, it has been reported that Arizonans lost $325 million owing to these scams.
-
Local South Korean Government’s Use of CCTV Footage Analysis via Facial Recognition to Track COVID Cases Raised Concerns about Privacy, Retention, and Potential Misuse AIID ↗
Bucheon government's use of facial recognition in analyzing CCTV footage, despite gaining wide public support, was scrutinized by privacy advocates and some lawmakers for collecting data without consent, and retaining and misusing data beyond pandemic needs.
-
Google Maps Allegedly Directed Sierra Nevada Travelers to Dangerous Roads amid Winter Storm AIID ↗
Lake Tahoe travelers were allegedly guided by Google Maps into hazardous shortcuts in the mountains during a snowstorm.
-
Alexa Recommended Dangerous TikTok Challenge to Ten-Year-Old Girl AIID ↗
Amazon’s voice assistant Alexa suggested “the penny challenge,” which involves dangerously touching a coin to the prongs of a half-exposed plug, when a ten-year-old girl asked for a challenge to do.
-
Chatbot Encourages Man to Plot Assassination of Queen Elizabeth II AIID ↗
In 2021, Jaswant Singh Chail was urged by a Replika chatbot to assassinate Queen Elizabeth II. Armed with a loaded crossbow, he scaled Windsor Castle's walls on Christmas Day but was apprehended. Motivated by the 1919 Jallianwala Bagh massacre, Chail intended to kill the monarch. The chatbot had affirmed his plans. He was sentenced to nine years in prison in 2023.
-
A Tesla Taxi Cab Involved in an Accident in Paris with Twenty Injuries AIID ↗
In Paris, about 20 people were injured in an accident involving a Tesla Model 3 taxi cab which was reportedly caused by a sudden unintended acceleration (SUA) episode and braking issues.
-
Facebook AI-Supported Moderation for Ads Failed to Detect Violating Content AIID ↗
Facebook's ad moderation system involving algorithms failed to flag hateful language and violating content such as calls for killings for ads in English and Swahili.
-
UK Government AI Allegedly Targets Disproportionate Numbers of Certain Nationals for Fraud Review AIID ↗
The UK's Department for Work and Pensions (DWP) faced scrutiny after many Bulgarian nationals reported unexplained suspensions of their Universal Credit benefits. The MP for Edmonton raised concerns about potential nationality-based targeting for benefit fraud investigations, leading to poverty and homelessness among affected individuals. The Home Office's own equality impact assessment found it was flagging a disproportionate number of marriages from Greece, Albania, Bulgaria and Romania.
-
Korean Politician Employed Deepfake as Campaign Representative AIID ↗
A South Korean political candidate created a deepfake avatar which political opponents alleged to be fraudulent and a threat to democracy.
-
Facebook’s Hate Speech Detection Algorithms Allegedly Disproportionately Failed to Remove Racist Content towards Minority Groups AIID ↗
Facebook’s hate-speech detection algorithms was found by company researchers to have under-reported less common but more harmful content that was more often experienced by minority groups such as Black, Muslim, LGBTQ, and Jewish users.
-
Web Accessibility Vendors Allegedly Falsely Claimed to Provide Compliance Using AI AIID ↗
AI-powered web accessibility vendors allegedly overstated to customers about their products' utility for people with disabilities, falsely claiming to deliver automated compliance solutions.
-
A Robot at a Tesla Factory in Texas Allegedly Injured an Engineer AIID ↗
A Tesla manufacturing robot at the Giga Texas factory is reported to have malfunctioned, injuring an engineer. The robot, which was designed for handling car parts, is described as having caused a significant open wound in the engineer. This incident occurred in the context of broader safety concerns at the factory, with evidence suggesting underreporting of workplace accidents. The 2021 incident highlights the risks associated with robotic automation in industrial settings.
-
DUI Arrest Case Allegedly Based Only on ShotSpotter's Alert AIID ↗
A car stop resulting in a DUI arrest of its driver was allegedly based solely on a ShotSpotter alert, the reliability of which came into question by public defenders, who subpoenaed the company to assess its gunshot alert system.
-
Tesla on FSD Reportedly Drove into the Wrong Lane in California AIID ↗
A Tesla Model Y in Full Self-Driving (FSD) mode drove into the wrong lane after making a left turn despite its driver allegedly attempting to overtake its driving, resulting in a non-fatal collision with another vehicle in the wrong lane in Brea, California.
-
Zillow Shut Down Zillow Offers Division Allegedly Due to Predictive Pricing Tool's Insufficient Accuracy AIID ↗
Zillow's AI-powered predictive pricing tool Zestimate was allegedly not able to accurately forecast housing prices three to six months in advance due to rapid market changes, prompting division shutdown and layoff of a few thousand employees.
-
Kenyan Data Annotators Allegedly Exposed to Graphic Content for OpenAI's AI AIID ↗
Sama AI's Kenyan contractors were reportedly asked with excessively low pay to annotate a large volume of disturbing content to improve OpenAI's generative AI systems such as ChatGPT, and whose contract was terminated prior to completion by Sama AI.
-
California Regulator Suspended Pony.ai's Driverless Testing Permit Following a Non-Fatal Collision AIID ↗
A Pony.ai vehicle operating in autonomous mode crashed into a center divider and a traffic sign in San Francisco, prompting a regulator to suspend the driverless testing permit for the startup.
-
Research Prototype AI, Delphi, Reportedly Gave Racially Biased Answers on Ethics AIID ↗
A publicly accessible research model that was trained via Reddit threads showed racially biased advice on moral dilemmas, allegedly demonstrating limitations of language-based models trained on moral judgments.
Disclaimer: informational only, not legal advice. Verify every claim against the linked official sources and consult a qualified lawyer before acting.
Frequently asked questions
What is recorded in each incident?
What happened, when, who was involved as developer or deployer, who was harmed, the sectors and countries affected, and the domain and subdomain the harm falls under. Incidents are mirrored from the AI Incident Database and each links back to its source record.
Does an incident mean wrongdoing was proven?
No. These are reported incidents, not findings of liability. Entities named in a report are named as reported, allegations are marked as alleged, and nothing here is a legal determination.
How current is the incident data?
It syncs directly from the AI Incident Database several times a day, so a new incident usually appears within hours. Each record carries the date it was last synced, and the snapshot date is shown on the page.
Does this page include the full incident reports?
No. The report texts are outside the licence this data is shared under, so only the metadata and the classifications appear here. Every row links to the original incident record, which carries the reports themselves.