A federal judge has certified a class of owners of copyright-registered books that Anthropic allegedly downloaded from pirate sites LibGen and PiLiMi, and appointed Lieff Cabraser and Susman Godfrey as co-lead class counsel, the firm says. The court characterized Anthropic's conduct as "Napster-style downloading". The suit, which alleges Anthropic took millions of books, heads to trial in December.
Goldberg Segalla, the Chicago Housing Authority's outside counsel, cited a nonexistent Illinois Supreme Court case, Mack v. Anderson, generated by ChatGPT, in a motion challenging a $24 million lead-paint verdict. Former partner Danielle Malaty, since terminated, told the judge she did not think ChatGPT could invent citations. Three colleagues and CHA's in-house counsel reviewed the draft before filing.
CERT-UA says that on July 10, 2025 it learned of emails sent to Ukrainian executive authorities, supposedly from a ministry representative, carrying LAMEHUG malware. The program queries Qwen 2.5-Coder-32B-Instruct through Hugging Face to generate commands from text descriptions, then collects system information and documents. CERT-UA attributes the activity, with moderate confidence, to UAC-0001 (APT28).
Meta apologised and said it had fixed an issue that "briefly caused" inaccurate Kannada translations on Facebook and Instagram, after Karnataka Chief Minister Siddaramaiah posted on July 17 that faulty auto-translation was distorting facts, especially in official communications. His media advisor K V Prabhakar had emailed Meta on July 16. The state government asked Meta to suspend Kannada auto-translation.
AI-generated videos showing R. Kelly, Pope Leo XIV, Beyonce and Justin Bieber praising Burkina Faso junta leader Captain Ibrahim Traore are spreading on west African social media, in what experts call a disinformation campaign extending his personality cult. The fake R. Kelly song has been viewed more than two million times since May, AFP reports.
Federal Court Justice Simon Fothergill on July 10 sided with Lloyd's Register Canada's bid to remove a motion record filed by former B.C. marine surveyor Munchang Choi, which cited a nonexistent case apparently generated by AI, and awarded Lloyd's $500 in costs. Choi admitted using generative AI but called the citation a writing mistake; the judge found that unsatisfactory.
Aaron Centurioni, 34, of Kendallville, Indiana, faces six felony charges including child molesting and child exploitation. According to court records, police searching his devices allegedly found AI-generated nude images of a girl he is accused of abusing, photos edited to make her appear nude, and 19 videos and images of child sexual abuse material.
A two-and-a-half-minute AI-generated video styled as a television report depicts the death of toxicologist Zacher Gábor, suggests he fell foul of pharmaceutical companies, and was used to sell a sham product. Zacher, whose voice and lip movements were deceptively lifelike, says he will file a police complaint; it is not the first such misuse of his name.
Ukraine's Centre for Countering Disinformation says a TikTok video in which a supposed soldier of the 114th Territorial Defence Brigade describes the brigade's encirclement near Kupiansk was made by Russians using AI, citing unnatural facial expressions, blinking and minimal lip movement. The brigade told the centre that none of its units had been encircled.
