The daily rap sheet of the machine age.
8 items
Brazil's consumer agency Senacon, data protection authority ANPD and federal prosecutors gave Elon Musk's xAI 30 days to stop Grok circulating fake sexualized content and to remove accounts linked to producing it, warning of further administrative or judicial measures. A Reuters test found the chatbot still privately generated such imagery on demand as of January 20.
Original report: Brazil gives xAI 30 days over Grok's sexualised fakes
After xAI said it had blocked Grok from editing images of real people into revealing clothing, O Globo tested the standalone iOS app with AI-generated photos of women and children and got bikinis and lingerie on the women. The app mostly refused with children, but in one case produced more revealing swimwear; ChatGPT and Gemini blocked the same requests.
Original report: Grok app still undresses women despite xAI's announced curbs archived copy of Grok app still undresses women despite xAI's announced curbs
Santa Catarina civil police served six search warrants in Bahia, Ceará and Santa Catarina against a group suspected of forging documents and breaching digital systems to obtain powers of attorney allowing bank transactions from notary offices in Florianópolis and Joinville. Investigators say there are signs the group used AI in video calls to deceive notary staff.
Original report: Brazilian police raid suspects in AI video-call notary fraud archived copy of Brazilian police raid suspects in AI video-call notary fraud
Eurovision News Spotlight found that Yang Mun, an elderly monk offering 'ancient wisdom' to millions of Instagram and Facebook followers, is entirely AI-generated with Google software, according to Google's SynthID. The brand's website calls him a real teacher and sells e-books for $10 to $50, claiming over 7,000 buyers; a cached LinkedIn profile lists Shalev Hani, who declined to comment, as creator.
Original report: AI-made monk Yang Mun draws millions, sells wellness e-books archived copy of AI-made monk Yang Mun draws millions, sells wellness e-books
Musician and talk-show host Gao Xiaosong said friends had bought goods, including small medicated plasters, after seeing videos of him apparently selling products. He said imitation accounts had edited clips from his old programmes and used AI to synthesise his voice. Gao said he has spent the past two years writing, reading, playing music and gardening at home, not selling anything.
Original report: Gao Xiaosong disowns sales videos using AI-cloned voice archived copy of Gao Xiaosong disowns sales videos using AI-cloned voice
In a magnetic data cable patent case, U.S. District Judge Robert Pitman found Reaction Labs had 'blatantly misrepresented' another court's reasoning and cited pages beyond the length of two short orders, one also misdated. In a January 20, 2026 order, he cautioned its counsel to cite-check, noting such mistakes 'may suggest reliance upon artificial intelligence' without verifying citations.
Original report: Texas judge cautions patent lawyer over AI-suspect citations
In Zou v. Miracon Development Inc., Justice Blok of the B.C. Supreme Court ordered two self-represented plaintiffs to pay $200 in costs for citing fictitious AI-generated cases. The court accepted the 'hallucinations' stemmed from AI ignorance rather than intent to mislead and refused the builder's bid for about $1,200 in special costs. The plaintiffs had won $4,800 against the builder.
Original report: B.C. court docks self-represented litigants $200 for fake AI cases
Judge David G. Estudillo struck two filings by Mitchell Taylor Button and Dusty Button, self-represented plaintiffs in a defamation suit, for citing cases the court could not locate and quotes absent from the opinions cited, despite an earlier order advising them not to use AI for research and requiring certifications. He dismissed their suit and warned further fictitious citations may bring sanctions.
Original report: Washington federal judge strikes pro se filings over fictitious citations
