The daily rap sheet of the machine age.
13 items
xAI has sued Terry Wayne Harwood, alleging he bypassed Grok's safeguards to turn non-sexual photographs into sexually explicit images without consent and to generate and distribute CSAM. Harwood, arrested in February, faces eight felony charges. xAI seeks damages for reputational harm and legal costs, and wants him barred from using Grok.
Original report: xAI sues South Carolina man over alleged Grok-made CSAM
According to data a hacker shared with 404 Media, Suno scraped millions of songs and lyrics from YouTube Music, Deezer and Genius, plus stock libraries including Pond5 and Jamendo. One file logged "2,013,545 music clips" from YouTube Music. The hacker says they also accessed user data for hundreds of thousands of customers and Stripe payment information.
Common Sense Media gave Google's AI Overviews and AI Mode its lowest rating after tests on accounts set up for 11- and 15-year-olds found they missed signs of suicidal ideation, offered deepfake instructions and recommended a helpline disconnected since 2023. Parents cannot switch the features off. Google disputes the findings and says it could not recreate them.
Original report: Google's AI search gets lowest kid-safety rating
Minas Gerais Civil Police are investigating, under seal, AI-generated fake intimate images of underage students at Colégio Genoma, a private school in Governador Valadares. Four students of the school allegedly shared the images in a private messaging group. The school opened an internal inquiry and preventively suspended the students involved.
Original report: Brazilian police investigate students over AI fake nudes of classmates
Busan police referred a YouTuber in his 40s to prosecutors, alleging he posted 16 AI-made fake videos from March to June. Police say he kept posting while under investigation, including a claim that masked Chinese police were assaulting people at Jamsil. They say he presented AI answers as FBI or CSI material, even after the AI warned they were unconfirmed. He denies the charge.
Original report: YouTuber passed off AI output as FBI files, police say
The High Court of Justice of Galicia fined a lawyer €1,800 over an AI-assisted appeal with 24 "inconsistencies or inventions". The court called filing such a "gross hallucination" a lack of respect and referred its ruling to the A Coruña bar association. The lawyer admitted the facts and apologised, but did not say which AI tool was used.
Original report: Galician court fines lawyer €1,800 over 24 AI-made errors
The court asked for a copy of "Bowers v. Transp. Sec. Admin.", cited in the government's motion to dismiss. An Assistant US Attorney in the Eastern District of North Carolina replied that the case does not exist and apologised. She said it likely came from a Google search returning AI-generated results. Her office began Westlaw cite-checking filings in March 2026.
Original report: Federal government lawyer admits citing a nonexistent case
During a 2 June urology consultation at Trofa Saúde Hospital da Trofa, a doctor handed a patient with a kidney problem a printout from Google's Gemini, labelled "AI overview", instead of a personalised nutrition plan, according to Correio da Manhã. The Medical Association said using AI for such basics is "profoundly inept". The hospital group did not respond to questions.
Original report: €90 consultation ends with a printout from Gemini
Hohhot's federation of literary and art circles revoked the first prize in a local photography contest after confirming that the winning image was AI-generated. It showed three sanitation workers on a bench, and netizens had spotted garbled text on their vests and matching faces. The event was suspended for review, and those accountable were held responsible.
Original report: Winning Hohhot photo's sanitation workers turn out to be AI
On 15 July, Romania's Border Police warned that social media accounts under the name "Cristina Anghel" were spreading AI-generated material using the force's visual identity. The force said the person associated with the accounts is not a border police officer and is not authorised to represent the institution. It urged citizens to rely on official channels.
Original report: Romanian Border Police disown AI-made 'Cristina Anghel' accounts
The Oregon Court of Appeals struck fabricated citations from a self-represented defendant's appeal brief. It also disregarded citations to real cases that did not support her arguments, denied her motion to file a corrected brief, and imposed a $100 fine. She said she did not know her drafting process could produce fabricated authority. Her convictions were affirmed.
Original report: Oregon appeals court fines appellant $100 for phantom citations
A W.D. Washington court struck pro se plaintiff Rebecca Jean's brief citing nonexistent authority, calling her explanations not credible. Asked about generative AI, Jean had listed "Microsoft Word to hand-type my response, and Google for research". She must now name any AI tool she uses and faces severe sanctions, up to and including dismissal, for future hallucinated citations.
Original report: Litigant names Microsoft Word as generative AI; court unconvinced
The Arizona Court of Appeals affirmed the probate of Jeffrey Kotchka's mother's will. It also sanctioned Kotchka for an opening brief containing hallucinated authorities obtained through generative AI, striking its inaccurate and false citations. Kotchka, who represented himself, said he relied on "artificial-intelligence-assisted research tools". The court noted that Arizona gives self-represented litigants no special leniency.
Original report: Arizona court sanctions pro se appellant for AI-hallucinated citations
