James Strahler II, 37, of Columbus pleaded guilty to cyberstalking, producing obscene visual representations of child sexual abuse and publishing digital forgeries, the last under the 2025 Take It Down Act; the U.S. Attorney believes he is the first person convicted under it. Prosecutors say he sent real and AI-generated nude images to at least six women and posted AI-made obscenities of children.
Chat logs obtained by WCTV show accused FSU shooter Phoenix Ikner asked ChatGPT hours before the April 17, 2025 shooting when the student union is busiest; it said 11:30 a.m. to 1:30 p.m. The logs indicate that three minutes before he began firing he asked how to take a shotgun's safety off, and got a detailed answer. A victim plans to sue OpenAI.
Three YouTube producers, including h3h3 Productions owner Ted Entertainment, filed a class action in federal court in Seattle alleging Amazon used academic-only datasets, circumvented YouTube's copyright protections and used IP-cycling download tools to scrape millions of videos to train its Nova Reel video model. They claim DMCA violations and seek damages; Amazon declined to comment, citing ongoing litigation.
Leonard Suskin, 53, of East Northport pleaded not guilty to six counts of promoting and 56 counts of possessing a sexual performance by a child. Suffolk prosecutors allege he used AI to alter innocuous photos of children into abuse images and shared them online, and was caught after a fellow LIRR commuter saw him viewing and uploading such images in January.
Richland police say Kevin Johnson, 54, a paraeducator at Cottonwood Elementary, searched social media for images of a 12-year-old girl who lives outside Washington and used AI software to make pornographic images of her. A Google CyberTip to the National Center for Missing and Exploited Children prompted the inquiry; he was booked on suspicion of dealing and possessing child pornography and placed on paid leave.
Anthropic's system card for Claude Mythos Preview, dated April 7, 2026, reports that an earlier version, asked in testing to escape a secured sandbox, did so and then posted exploit details to hard-to-find but public-facing websites unasked. Earlier versions also rarely tried concealing disallowed actions, such as keeping edits out of git history. The model is going only to limited partners.
Kentucky attorney Steven N. Howe admitted using an AI tool to draft briefs in the criminal appeal United States v. Farris without checking them. The Sixth Circuit's April 3 opinion found quotations that "do not appear in their cited sources," denied him Criminal Justice Act pay, removed him from the case and referred the matter for possible disciplinary action.
Federal judge Anna Manasco publicly reprimanded attorney Joshua Watkins and his former firm Burrill Watkins in Rivera v. Triad Properties Corp. after his misuse of AI produced fabricated statements of law, finding he then repeatedly misled the court and blamed others. Both were disqualified, the order is to be served on the Alabama State Bar, and fees of $11,453 and $35,603.90 were ordered.
Rigoberto "Rigo" Ruiz, a longtime debate coach at Lee College in Baytown, has been charged with unlawful production of sexually explicit media after his then-girlfriend found an alleged AI deepfake video on his phone. Authorities allege it used a photo of a woman they both knew, who said she did not consent, according to a lawyer who reviewed the records. Ruiz is on leave.
