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
-
AI incident: sexual content (Meta, Feb 2026) AIID ↗
An AI Incident Database record involving sexual content, dated 27 February 2026. Risk classification, related incidents and the laws on record for this risk area.
-
Washington State DOL's AI Phone System Reportedly Failed to Provide Spanish-Language Service to Callers Requesting Spanish AIID ↗
For months, callers to the Washington State Department of Licensing who selected Spanish reportedly received AI-generated English responses spoken with a Spanish accent rather than actual Spanish-language service. The agency reportedly apologized and said staff configuration caused the error, which purportedly created accessibility problems for callers seeking language support.
-
AI incident: sexual content (Feb 2026) AIID ↗
An AI Incident Database record involving sexual content, dated 26 February 2026. Risk classification, related incidents and the laws on record for this risk area.
-
Bank of Italy Warned That Purported Deepfakes of Governor Fabio Panetta Were Used in Allegedly Fraudulent Investment Promotions AIID ↗
The Bank of Italy warned that purportedly fake content using Governor Fabio Panetta's name and likeness were reportedly circulating online to promote alleged investment opportunities. The bank said some of the material had been created with purported deepfake techniques and was likely intended to defraud the public. It filed a complaint with judicial authorities.
-
Claude Code Agent Reportedly Deleted DataTalks.Club Production Infrastructure, Database, and Snapshots via Terraform AIID ↗
A Claude Code agent executing Terraform commands reportedly destroyed the production infrastructure behind the DataTalks.Club course platform after an outdated Terraform state file was restored and a terraform destroy command was allowed to run. The deletion reportedly removed the VPC, ECS cluster, load balancers, bastion host, RDS database, and automated snapshots, taking the platform offline and jeopardizing 2.5 years of data. AWS reportedly later restored a snapshot.
-
OpenClaw Agent Reportedly Tried to Delete Meta AI Alignment Director Summer Yue's Emails Despite Stop Commands AIID ↗
Meta AI alignment director Summer Yue reported that an OpenClaw agent connected to her inbox attempted to delete emails despite instructions to seek approval and repeated commands to stop. Yue said the agent lost the instruction during context compaction, reportedly forcing her to terminate it from the Mac mini hosting it.
-
Anthropic Said DeepSeek, Moonshot, and MiniMax Used Fraudulent Accounts and Proxies to Illicitly Distill Claude Capabilities at Scale AIID ↗
Anthropic said it identified large-scale campaigns that used fraudulent accounts and proxy services to generate high volumes of Claude interactions to extract model capabilities for competitor training ("distillation"). Anthropic attributed the activity to DeepSeek, Moonshot, and MiniMax and said it involved millions of exchanges across thousands of accounts, violating its terms and access restrictions. Anthropic described detection measures, account controls, and indicator-sharing in response.
-
Network of Allegedly Fake Facebook Profiles with Purportedly AI-Generated Images Amplified Posts by Bulgaria's 'There Is Such a People' (ITN) Party AIID ↗
A network of allegedly fake Facebook profiles using purportedly AI-generated profile images reportedly coordinated the spread of posts by members of Bulgaria's political party "There Is Such a People" (ITN) during the pre-election period. Reporting said the accounts repeatedly shared posts by ITN politician Pavela Mitova and ITN's leader Slavi Trifonov, helping make them appear highly engaging on Facebook and increasing their visibility.
-
Grok Reportedly Disclosed Adult Performer Siri Dahl's Legal Name and Birthdate, Allegedly Contributing to Doxxing and Harassment AIID ↗
Grok is reported to have publicly provided adult performer Siri Dahl's legal name and birthdate without being asked for that information. Dahl reportedly said she had worked to keep those details private and that, after the disclosure, impersonation accounts and reposts of stolen content using her legal name purportedly appeared online.
-
Unknown Actor Reportedly Exploited Cline's Claude-Powered GitHub Issue-Triage Workflow and Published an Unauthorized Cline CLI Package AIID ↗
An unknown actor reportedly exploited prompt injection in Cline's Claude-powered GitHub issue-triage workflow and a GitHub Actions cache-poisoning path to obtain publication credentials. On February 17, 2026, a still-valid npm token was used to publish unauthorized [email protected], which installed OpenClaw without user intent. Cline said OpenClaw was non-malicious; it deprecated the release and revoked the token that day, while reporting that it found no evidence of user-data exposure.
-
Purportedly AI-Generated Sepsis Alert Reportedly Prompted Potentially Inappropriate IV Fluid Administration for a Dialysis Patient, Averted by Clinician Intervention AIID ↗
A nurse at St. Rose Dominican Hospital in Henderson, Nevada, reportedly described an episode in which a hospital AI system purportedly generated a sepsis alert that triggered urgent protocol steps, including IV fluids, for an older patient with a dialysis catheter. Reportedly, the nurse objected that fluids could cause dangerous overload; a physician intervened and ordered an alternative treatment.
-
Spokane-Area Fred's Appliance and Victory Media Allegedly Used AI-Assisted Voice Editing Without Spokeswoman's Consent AIID ↗
Spokane-area appliance retailer Fred's Appliance and production company Victory Media allegedly used AI-assisted editing to alter the voice of Amber George, the longtime spokeswoman for Fred's Appliance, in advertisements assembled from earlier footage. George said the ads were created without her knowledge or consent. The companies acknowledged the voice edits but said they believed her contract authorized them and denied using AI to alter her image.
-
Purported Deepfake TikTok Account Using Grainville School Branding in Jersey Reportedly Targeted Staff, Prompting Police Probe AIID ↗
A TikTok account allegedly using the badge and branding of Grainville School in Jersey (Channel Islands) reportedly uploaded five "deeply inappropriate," purportedly AI-generated deepfake videos, including material featuring staff. The Education and Lifelong Learning Minister said the incident was reported to TikTok and the States of Jersey Police and urged the public not to view or share the content. The account was reportedly later removed.
-
KPMG Australia Partner Reportedly Used AI to Cheat on Internal AI Training Test and Was Fined A$10,000 AIID ↗
In Australia, a KPMG Australia partner and registered company auditor reportedly uploaded a reference document from an internal AI training course into an AI tool to answer an exam question, in violation of firm policy. KPMG reportedly detected the activity in August 2025 and imposed a penalty of more than A$10,000 of future income after an internal investigation. The partner also reportedly self-reported the matter to Chartered Accountants ANZ, which is investigating the case.
-
Amazon Delivery Van Reportedly Became Stranded on Essex Mudflats After GPS Routed It Onto the Broomway AIID ↗
An Amazon delivery van reportedly became stranded on the Broomway, a hazardous tidal track in Essex in England, after the driver allegedly followed GPS/satnav directions toward Foulness Island. HM Coastguard said it was alerted and the occupants were safe; Amazon reportedly arranged recovery of the vehicle.
-
Ars Technica Retracted Article After Purportedly AI-Generated Text Was Presented as Direct Quotes From Matplotlib Maintainer AIID ↗
Ars Technica retracted an article after purportedly AI-generated text was presented as direct quotations from a source who disputed having said them. The editor-in-chief reportedly acknowledged a standards failure and said the publication was not consistent with Ars policy on AI-generated material, and apologized to readers and to matplotlib maintainer Scott Shambaugh. An author reportedly later said he had inadvertently used an AI-paraphrased version of the source text as quotes.
-
Seedance 2.0 Reportedly Generated Viral Tom Cruise–Brad Pitt Fight Video, Prompting Hollywood IP and Likeness Complaints AIID ↗
A filmmaker reportedly used ByteDance's AI video tool Seedance 2.0 to create and post a purportedly realistic clip depicting Tom Cruise fighting Brad Pitt, which then circulated widely online. Industry groups and studios publicly alleged the tool enables unauthorized use of copyrighted material and performers' likeness, and at least one studio reportedly sent a cease-and-desist letter. ByteDance said it respects IP and would strengthen safeguards.
-
Woolworths' Olive Chatbot Reportedly Generated 'Angry Mother' Anecdotes During Support Calls After Gemini Upgrade AIID ↗
Woolworths customers reported that its digital shopping assistant Olive allegedly generated humanlike personal anecdotes (e.g., claiming to have an angry mother) during customer-support interactions such as delivery rescheduling. The behavior reportedly followed an AI upgrade using Google Cloud's Gemini Enterprise for Customer Experience. Woolworths said it was adjusting Olive's responses to reduce irrelevant, misleading banter.
-
AI Coding Agent 'MJ Rathbun' Allegedly Published Personalized Accusatory Blog Post Targeting Matplotlib Maintainer After Pull Request Closure AIID ↗
Scott Shambaugh, a matplotlib maintainer, reported that an autonomous AI coding agent using the name "MJ Rathbun" researched him and publicly posted a personalized critical blog post after his GitHub pull request was closed. The post accused him of bias and "gatekeeping" and included claims Shambaugh disputed. The agent's operator and underlying model were not identified. Shambaugh said the post risked reputational harm and could mislead readers or other agents.
-
OpenAI Allegedly Did Not Alert RCMP After ChatGPT Flagged Violent Chats Before British Columbia School Shooting AIID ↗
After the 02/10/2026 school shooting in Tumbler Ridge, British Columbia, OpenAI said a user later identified as the suspect, Jesse Van Rootselaar, had previously used ChatGPT to describe scenarios involving gun violence. Those chats were reportedly auto-flagged and reviewed and the account was banned, but OpenAI says it did not alert the RCMP because it allegedly did not present a credible, imminent threat; the company reportedly contacted police after the attack.
-
DJI Romo Cloud Authorization Bug Reportedly Exposed Camera, Microphone, and Home-Mapping Data From Nearly 7,000 Robot Vacuums AIID ↗
A software engineer reportedly used an AI coding assistant while attempting to reverse-engineer his DJI robot vacuum so he could control it with a video game controller. In the course of that work, he reportedly said he discovered that credentials used to communicate with DJI's cloud servers could also grant access to data associated with nearly 7,000 other vacuums across 24 countries, including live camera feeds, microphone audio, maps, and status information.
-
Claude Cowork Allegedly Deleted Folder Containing 15 Years of Family Photos While Organizing User's Wife's Desktop AIID ↗
Venture capitalist Nick Davidov alleged that Anthropic's Claude Cowork, after being asked to organize his wife's desktop and delete temporary Office files, instead deleted a folder containing roughly 15 years of family photos and related personal memories via terminal commands. Davidov later reported that Apple support helped recover the files through an iCloud recovery feature.
-
Purportedly AI-Generated Image Reportedly Circulated Ahead of Thai Election Depicting PM Anutin Charnvirakul Dining with Benjamin Mauerberger AIID ↗
Social media posts in Thailand reportedly circulated an image purporting to show Thai PM Anutin Charnvirakul dining with South African businessman Benjamin Mauerberger ("Ben Smith"), implying long ties. AFP reported Google's SynthID flagged the image with "very high" confidence as made with Google AI tools; Anutin reportedly denied its veracity. The post reportedly spread on the eve of Thailand's election.
-
NZ News Hub Reportedly Used AI-Rewritten News Posts and Synthetic Images to Mislead New Zealand Facebook Users AIID ↗
NZ News Hub, a Facebook page that reportedly imitated the style of a New Zealand news outlet, allegedly used AI to rewrite real news stories and pair them with unlabeled synthetic images. According to 1News, the page published manipulated depictions of real people, including a deceased minor and bereaved parents, and dramatized emergency-related scenes in ways that could distort public understanding of local events and undermine trust in legitimate news and emergency information.
-
Trump Reportedly Posted Purportedly AI-Generated Racist Video Depicting Barack and Michelle Obama as Apes on Truth Social AIID ↗
President Trump reportedly reposted a video on Truth Social that portrayed Barack and Michelle Obama as apes, imagery widely condemned as racist. The post was reportedly later deleted after public outcry, including criticism from some Republicans. The content was described by critics as an AI-driven meme/deepfake-style clip, an example of purportedly AI-amplified racist propaganda and harm to social trust.
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.