AI Incidents
Real-world cases where AI systems caused harm, classified by risk domain and entity. Data sourced from 
AIID and classified by 
MIT AI Risk Initiative. CC BY-SA 4.0
58 incidents
Judge Reportedly Disqualifies Butler Snow Lawyers Following Purported Use of ChatGPT-Fabricated Citations in Alabama Prison Litigation
Three Butler Snow attorneys defending a former Alabama corrections commissioner were reportedly disqualified after two motions cited five non-existent cases generated via ChatGPT. Judge Anna Manasco is reported to have publicly reprimanded the lawyers, referred them to the Alabama State Bar, and ordered broad disclosure; she declined to sanction the firm. An external review of 2,400 citations across 330 filings reported no further errors.
Meta AI App Reportedly Publishes Personal Chats Without Users Fully Realizing
Meta launched a stand-alone AI app with a "Discover" feed allowing users to share conversations with its chatbot. Multiple reports indicate that some users may have inadvertently published highly personal interactions, including audio recordings, medical questions, legal concerns, and intimate relationship disclosures. While Meta states that sharing is opt-in, the feature's design and labeling may have led to user confusion about what would be publicly visible.
Factum in Ko v. Li Allegedly Contains AI-Generated Case Law Citations
In Ko v. Li, 2025 ONSC 2766, the Ontario Superior Court of Justice identified multiple inaccurate or nonexistent case citations in a factum submitted by counsel for the applicant. Justice FL Myers questioned whether generative AI was used in drafting the document, noting the cited cases were either irrelevant, misrepresented, or did not exist. The court ordered the lawyer to show cause why she should not be cited for contempt, citing duties of accuracy and technological competence.
$31,000 Sanction in Lacey v. State Farm Tied to Purportedly Undisclosed Use of LLMs and Erroneous Citations
In the case of Lacey v. State Farm, two law firms were sanctioned $31,000 after submitting a legal brief containing reportedly erroneous citations generated using AI tools. The court reportedly found that the lawyers failed to disclose the use of AI, neglected to verify its output, and refiled a revised brief with additional inaccuracies. Judge Michael Wilner deemed the conduct reckless and issued sanctions for what he described as "improper" and "misleading" legal filings.
Reportedly Unsafe Deployment of Llama.cpp Reveals Interactive AI-Generated CSAM Roleplay Prompts
A study by UpGuard reports that misconfigured llama.cpp servers publicly exposed user prompts, including hundreds of interactive roleplay scenarios. Some prompts explicitly described fictional sexual abuse of children aged 7–12. While no real children were involved, the findings demonstrate how open-source LLMs can be exploited to generate AI-enabled child sexual abuse material (CSAM).
Norwegian Supreme Court Receives Legal Filing with Fabricated Citations Allegedly Generated by AI Tool
In early 2025, Norway’s Supreme Court (Høyesterett) received a legal filing containing hallucinated legal citations and quotes, allegedly inserted by an attorney using an AI tool without verification. The fabricated sources were detected during standard review. While no sanctions were issued, the Court updated its guidance to lawyers, warning that AI use does not exempt them from professional responsibility.
GenNomis AI Database Reportedly Exposes Nearly 100,000 Deepfake and Nudify Images in Public Breach
In March 2025, cybersecurity researcher Jeremiah Fowler discovered an unprotected database linked to GenNomis by AI-NOMIS, a South Korean company offering face-swapping and "nudify" AI services. The exposed 47.8GB dataset included nearly 100,000 files. Many depicted explicit deepfake images, some involving minors or celebrities. No personal data was found, but the breach was a serious failure in data security and consent safeguards in AI image-generation platforms.
Alleged ChatGPT Misuse by Contractor Leads to Reported Data Exposure in New South Wales Resilient Homes Program
A former contractor of the New South Wales Reconstruction Authority reportedly uploaded a spreadsheet containing personal and health information of Resilient Homes Program applicants to ChatGPT during a three-day period in March 2025. Up to 3,000 people may have reportedly been affected.
Sydney Students Allegedly Forced to Retake NAPLAN After AI Predictive Text Error
A predictive text malfunction allegedly compromised the integrity of the NAPLAN writing exam at two Sydney schools, Waverley College and Kambala, in that it allowed students to access AI-driven text suggestions. Caused by a technical oversight, this incident led to affected students being required to retake the test.
Alleged AI-Generated Video by Spain’s People’s Party Results in Diplomatic Fallout with the Dominican Republic
Spain's People's Party (PP) posted an AI-generated attack video depicting Prime Minister Pedro Sánchez on a beach under the title "The Island of Corruption," a reference to a reality TV show filmed in the Dominican Republic, indirectly linking the country to corruption. The Dominican Foreign Ministry condemned the video as a "vicious attack" for using its national symbols. PP later deleted the post, and Sánchez apologized on behalf of Spain.
Alleged Inclusion of 12,000 Live API Keys in LLM Training Data Reportedly Poses Security Risks
A dataset used to train large language models allegedly contained 12,000 live API keys and authentication credentials. Some of these were reportedly still active and allowed unauthorized access. Truffle Security found these secrets in a December 2024 Common Crawl archive, which spans 250 billion web pages. The affected credentials could have been exploited for unauthorized data access, service disruptions, financial fraud, and a variety of other malicious uses.
MyPillow Defense Lawyers in Coomer v. Lindell Reportedly Sanctioned for Filing Court Document Allegedly Containing AI-Generated Legal Citations
In February 2025, lawyers Christopher I. Kachouroff and Jennifer T. DeMaster, representing Mike Lindell, reportedly used generative AI to draft a court brief that contained nearly 30 defective or fabricated citations. The error-filled filing violated federal court rules requiring factual and legal accuracy. The judge fined both lawyers $3,000 each, citing either the improper use of AI or gross carelessness as the cause of the misleading legal content.
ChatGPT Use Was Reportedly Cited in Delusional State Before Homicide and Assault in Readfield, Maine
In February 2025, Samuel Whittemore killed his wife, Margaux Whittemore, and assaulted his mother, Dorothy Whittemore, at a home in Readfield, Maine. Court reporting said forensic psychologists testified he was experiencing bipolar disorder with psychotic features and delusions, and that he had used ChatGPT up to 14 hours a day as a companion before the attack. Current reporting links ChatGPT use to his deteriorating mental state but does not establish direct violent instruction.
Purportedly AI-Assisted Report by Tromsø Officials Allegedly Cited Non-Existent Sources in School Closure Proposal
In February 2025, Tromsø Municipality reportedly presented a report proposing the closure of schools and kindergartens. Residents reportedly later discovered the report cited numerous non-existent sources. Officials reportedly admitted using generative AI to help draft the report, leading to fabricated references. The incident reportedly led to public outcry and suspension of the closure plan. It is purportedly Norway's first major public-sector AI scandal.
Alleged Misuse of Facial Recognition Technology by Law Enforcement Reportedly Leading to Wrongful Arrests and Violations of Investigative Standards
Law enforcement agencies across the U.S. have allegedly been misusing AI-powered facial recognition technology, leading to wrongful arrests and significant harm to at least eight individuals. Officers have reportedly been bypassing investigative standards, relying on uncorroborated AI matches to build cases, allegedly resulting in prolonged detentions, reputational damage, and personal trauma.