An Oklahoma woman spent six months behind bars after Maryland police relied on a facial recognition match sent back by someone on a police-and-investigator listserv, according to the ACLU. Charges in three counties were eventually dropped or dismissed. The ACLU says a detective concealed the technology from the court, and it is demanding apologies from three Maryland police departments.
The NAACP and its Mississippi State Conference have sued xAI and its subsidiary MZX Tech, alleging they installed and ran twenty-seven gas turbines in Southaven, Mississippi, without Clean Air Act permits to power the Colossus 2 data center behind Grok. The complaint says the turbines could emit about 1,700 tons of NOx a year near communities with large Black populations.
The State Bar of California has filed disciplinary charges against Los Angeles attorney Omid Emile Khalifeh and Scottsdale attorney Steven Thomas Romeyn, alleging their AI-assisted filings cited nonexistent cases; the State Bar Court has not ruled. Separately, on April 6 the court approved probation including a 30-day suspension for attorney Sepideh Ardestani, who blamed her erroneous citations on handwritten notes.
Chief Judge Richard Seeborg of the Northern District of California refused on Monday to stay or dismiss a suit against OpenAI and Sam Altman alleging a man's hundreds of hours of conversation with ChatGPT led him to kill his mother and himself. He ruled that a similar state-court case brought by the mother's estate was not sufficiently parallel.
An NBC News review found dozens of Grok-made sexualized images and videos of real women, many of them pop stars or actors, posted on X over the past month, despite X's January pledge to stop such edits; users dodged the restrictions with pose-matching and clothing-swap prompts. Most of the posts vanished after NBC asked; X says xAI has "extensive safeguards."
The Amsterdam Court of Appeal ruled on Tuesday that X must give Leiden PhD researcher Danny Mekić access to the "Guano Notes", the labels X attached to his account that left it temporarily unfindable in 2023, with only moderator names and exact times redacted. X faces €4,000 for each day of non-compliance and has said it may take the case to cassation.
After CCTV reported that users of Chinese creation platform LiblibAI could slip coded prompts past moderation to generate videos of half-naked women dancing, the company said on April 14 that an internal review found it had generated non-compliant content in some complex prompt combinations, apologised, and said the gaps were fixed. Beijing News testers found the prompts no longer worked.
Mato Grosso civil police launched Operation Mil Faces on Tuesday, serving 13 court orders, including two preventive arrest warrants, against a group suspected of using generative AI to fake facial biometrics at a telephone provider. Once fake identities passed, police say, the group took over victims' phone lines via SIM swaps, and hundreds of consumers across Brazil lost money or faced improper purchases.
Toyama police have arrested a 42-year-old Nagano Prefecture man accused of displaying obscene AI-generated images on a paid members-only site; investigators say he produced an estimated 40,000 images depicting fictional elementary and middle school-aged girls, drawing about 140,000 visits a month. He reportedly admitted the charges. It is Toyama police's first crackdown on AI-generated obscene material.
The Kassel Administrative Court ruled that undisclosed AI use in academic work constitutes academic dishonesty, after two students went to court over the University of Kassel's suspicion that a master's term paper and a bachelor's thesis were largely AI-written. Unlike search engines, AI supplies phrasing, structure and reasoning, the court said; even a single unattributed use can count, though AI spell-checking does not.
