AI Misconduct.com

The daily rap sheet of the machine age.

Updated Nov 2, 2026
4,173 items in the archive
RSS feed

Thursday, April 30, 2026

13 items

Home Affairs suspends two officials over AI-hallucinated White Paper references

Who did it: HYBRID South African Government

The Department of Home Affairs is suspending two officials after apparent AI "hallucinations" turned up in the reference list of its Cabinet-approved Revised White Paper on Citizenship, Immigration and Refugee Protection. The references were not cited in the body, the department says. It has withdrawn the list and appointed two law firms, one to review policy documents back to 30 November 2022.

Karlsruhe prosecutors charge man over AI-generated child abuse images

Who did it: HUMAN SWR Aktuell

Karlsruhe prosecutors and police have charged a 59-year-old man with using AI programs to create child sexual abuse images in four cases in November 2023, allegedly training the programs for results as realistic as possible. He is also accused of possessing such material from 2020 to 2024 and making it accessible over a network. A Karlsruhe court must decide whether to admit the charges.

Delhi High Court stays judgment flagged for possible AI drafting

Who did it: HYBRID AI Hallucination Cases database

Akasa Air operator SNV Aviation alleges a Saket commercial court judgment ordering it to pay Rs 1,08,80,000 to a tour operator attributed non-existent propositions of law to real Supreme Court cases. On April 30 the Delhi High Court said the drafting prima facie suggested AI software may have been used, stayed the judgment subject to a Rs 20 lakh deposit, and requisitioned the trial record.

Austin officials say up to five Waymos blocked shooting response

Who did it: AI Spectrum News 1 South Texas / El Paso

Austin public safety officials told a council committee meeting that as many as five Waymo robotaxis blocked the response corridor during the March 1 mass shooting outside Buford's. Firefighters reported several became "confused" and stopped at 6th and Guadalupe, the lead one not moving despite hand signals, and ambulances got caught in the backup. Waymo skipped the meeting, leaving its reserved seats empty.

Meta threatens New Mexico exit over child-safety demands

Who did it: HUMAN Fortune

After a Santa Fe jury in March 2026 found Meta liable for 75,000 violations of New Mexico's Unfair Practices Act and ordered $375 million in civil penalties, Attorney General Raúl Torrez filed for sweeping injunctive relief, including changes to recommendation algorithms for minors. Meta called the demands "impossible for any company to meet" and said it may remove its platforms from the state.

Hangzhou court rules firing a worker replaced by AI unlawful

Who did it: HUMAN 新华网

The Hangzhou Intermediate People's Court has upheld a ruling that an AI-related tech company unlawfully fired a quality assurance supervisor after large language models took over the role. The company had offered a lower-level post at 15,000 yuan a month, down from 25,000; the court said AI replacement was not a "major change in the objective circumstances" justifying termination.

Lawyer blames malware for fake citations, gets $1,500 sanction

Who did it: HYBRID AI Hallucination Cases database

An Eastern District of New York magistrate judge imposed a $1,500 sanction on attorney Harry R. Thomasson, a defendant representing himself, over a brief citing two non-existent cases the court called AI-generated. Thomasson said he does "not use AI" and blamed computer "viruses and malware", producing Staples paperwork; the court said he still failed to verify his citations.

Italy's AGCOM asks EU to investigate Google's AI search features

Who did it: HYBRID Reuters

Italy's communications watchdog AGCOM has referred Google Ireland Ltd to the European Commission for a formal assessment of its AI Overviews and AI Mode under the Digital Services Act, following a complaint from newspaper publishers' federation FIEG, which argues the AI summaries divert readers from original news sources and could spread "hallucinations". Google had no immediate comment.

Ohio judge orders lawyer to pay $5,273 over AI-fabricated citations

Who did it: HYBRID AI Hallucination Cases database

Judge Steven K. Dankof of the Montgomery County Common Pleas Court ordered attorney Ghassan Jamal Deek to pay Shell and True North Energy $5,273 in fees and expenses. At a March 13, 2026 hearing Deek admitted AI was used to produce every filing he submitted, including one citing a non-existent "Perrico v. Verizon"; earlier he had blamed another attorney and his own client.

Leon County deputies arrest man over AI-generated child abuse material

Who did it: HUMAN WCTV

Leon County deputies arrested an 18-year-old on Wednesday following CyberTips from electronic service providers to the National Center for Missing and Exploited Children; he faces several counts including possession of generated child pornography. Deputies say a forensic analysis of his seized devices found "AI-generated and actual child sexual abuse material, as well as animal sexual abuse material."

Lawsuit accuses Phoenix men of selling AI porn-influencer courses

Who did it: HYBRID Wired

Three plaintiffs sued Jackson Webb, Lucas Webb and Beau Schultz in Arizona, alleging the Phoenix men used photos of unsuspecting women to generate look-alike AI models, sold the content on Fanvue and charged $24.95 a month on Whop for courses teaching others to copy them, advising them to pick women with under 50,000 followers to avoid 'legal issues'. The men did not comment.

Flock used cameras in children's gymnastics room for sales demos

Who did it: HUMAN 404 Media

Access logs obtained by a Dunwoody, Georgia, resident show Flock sales staff tapped cameras in sensitive spots, including a children's gymnastics room and pool at the Marcus Jewish Community Center of Atlanta, to demo its surveillance tech to police elsewhere. Flock says the city authorised it under a demo partner program, denies spying on children, and will stick to places like retail parking lots.

Friends' School denies discouraging parents from telling deepfake victims

Who did it: HUMAN ABC News (Australia)

Two parents say staff at Tasmania's The Friends' School told them most families weren't telling daughters they had been identified in AI-generated pornographic images allegedly shared in a boys' group chat, and that police told them 21 girls were victims. Principal Esther Hill denies families were ever encouraged to withhold information; five boys have left and police said no charges have been laid.