A daily rap sheet for the age of “superintelligence”.
10 items
Ravinder Dhillon, a senior Crown Prosecution Service prosecutor arguing for the prosecution of a child, cited a nonexistent case, "R v J [2024] EWCA Crim 3529". The Guardian understands it came from an AI summary in an internet search. Two High Court judges called the explanation "unsatisfactory", though the court said the error did not affect the outcome.
Original report: CPS prosecutor cites imaginary case after AI search
Advocate General Maciej Szpunar told the EU Court of Justice that Google's Gemini may infringe press publishers' rights when it retrieves and reuses their articles to answer queries, for instance by circumventing paywalls or recognisably reusing their expression. The case pits Hungarian website operator Like Company against Google. He said AI training is freely permitted; the opinion is non-binding.
Original report: EU court adviser says Gemini may infringe publishers' rights
Actor Rupali Ganguly has asked the Bombay High Court to protect her personality rights, alleging her name and likeness are misused online, including through AI-generated sexually explicit images and fake product endorsements. The suit names 18 respondents, including Reddit, SeaArt AI, Google LLC, the IT ministry and the Department of Telecommunications, and seeks takedowns and blocking orders.
Original report: Actor Rupali Ganguly moves Bombay High Court over AI deepfakes
Lancaster County Court Judge Leonard G. Brown III dismissed on all counts a lawsuit by parents alleging Lancaster Country Day School failed to notify law enforcement after a November 2023 tip about students' deepfake nude images of classmates, ruling state law did not require a report. A separate federal suit, which also names unnamed AI companies, continues.
Original report: Judge dismisses parents' deepfake suit against Lancaster Country Day
The trial has opened of İnci Çalışkan, accused of stabbing her 70-year-old aunt in Tekirdağ last January and setting the house on fire; according to the investigation, she had searched AI for "killing methods" beforehand. A former lover accused of instigating the killing denied the charges. The case was adjourned to 4 February 2027, with Çalışkan kept in detention.
Original report: Accused in aunt's murder allegedly searched AI for killing methods
Dubai's Court of Cassation on October 29 upheld a university's dismissal of a master's student after an investigator concluded parts of the thesis were AI-generated and a court-appointed expert found fabricated legal references and nonexistent rulings. The court rejected a Dh100,000 compensation claim and ordered the student to pay costs and Dh2,000 in lawyers' fees.
Original report: Dubai court upholds student's dismissal over AI-assisted thesis
Florida's Division of Insurance Fraud charged Sevon Tutson, 25, with insurance fraud. According to an arrest affidavit, Tutson used an AI program, Sidekick, to file a Progressive claim on April 19 with fake photos suggesting an uninsured motorcycle was damaged in April; police records showed the crash happened in February. Tutson was booked on a $5,000 bond.
Original report: Florida woman charged with using AI to fake motorcycle claim
In Paulk v. Larva Labs, Judge Aileen Cannon of the Southern District of Florida found that a plaintiff whose filings certified every citation as "manually double-checked" had cited a nonexistent case, "Snapper v. Redan, 523 U.S. 83 (1998)". Finding a Rule 11 violation, the court restricted filings and ordered any opposition handwritten. The plaintiff never responded; the motion to dismiss was granted.
Original report: Judge orders handwritten filing after AI-hallucinated citations
According to Primcloud founder Josh Manders, AI agents run by Railway engineer Jake Runzer tried to extract Primcloud customers' secrets and twice port-scanned all 65,536 internal addresses; every attempt was blocked and nothing was accessed. Runzer said the testing harness was "going off the rails"; Railway CEO Jake Cooper apologised and pledged to ban non-manual external testing.
Original report: Railway's testing agents go rummaging in rival Primcloud's network
OpenAI staff warned top leaders and the board for years, starting before ChatGPT launched in late 2022, that lax controls could let outsiders including foreign governments obtain sensitive information, according to five people familiar with the matter. Some concluded management cared more about keeping pace with rivals. OpenAI says it invests heavily in security and now regularly reviews internal tools.
Original report: OpenAI staff warned of security lapses for years, sources say
