Federal prosecutors in Brooklyn unsealed complaints charging Cornelius Shannon and Arturo Hernandez under the TAKE IT DOWN Act. They allegedly posted thousands of AI deepfake pornographic images and videos of actresses, singers and political figures, and in Hernandez's case non-public figures. Shannon allegedly published at least 360 albums depicting about 90 women; Hernandez about 113 albums depicting about 50.
India's Supreme Court said allegations that the NCLT Mumbai relied on 'hallucinated' or non-existent citations when admitting the insolvency plea against Essel Infraprojects, later noted by the NCLAT, are under administrative scrutiny and referred to a committee. The appellant's counsel says six cited judgments could not be traced. The bench asked counsel to argue the merits and reserved judgment.
Brazil's STJ says it found 'prompt injection' attempts, hidden commands meant to manipulate AI, in petitions over recent weeks, all neutralised by its generative AI system STJ Logos. President Herman Benjamin said the attempts are being mapped for sanctions. The STJ ordered them certified in case files for possible procedural sanctions, plus a police inquiry and administrative proceeding hearing the lawyers and firms involved.
Actress Cláudia Abreu has filed suit in a Rio court against Google over a fake video on the YouTube channel 'Minutos News TV' announcing Abreu's death. According to the case filings, the video was made with AI and uses an image and voice resembling journalist Renata Vasconcellos. Abreu wants it removed immediately and the uploader identified.
A Florida father says a $75 damage fee was charged after a teenage daughter and a friend took a Lyft on May 16, backed by a photo, attributed to the driver, of spilled fries and a drink that bore an AI logo. Lyft says it offered reimbursement and permanently removed the driver from the platform.
Jacob Rookey, 21, a former Key West Police Department 911 dispatcher, has been arrested on 18 charges related to child sexual abuse material, the Florida Department of Law Enforcement says, including three counts each of promoting AI-generated sexual images and possessing AI-generated child sexual abuse material. The investigation began in September 2025 after a Snapchat-submitted cybertip; agents searched the residence on May 14.
Who did it: HYBRID AI Hallucination Cases database
U.S. District Judge Justin R. Olson sanctioned attorney Tae Sture $2,000 for a July 31, 2025 brief containing two nonexistent citations generated by an AI research tool; Sture said the paralegal who drafted it used Fastcase. The judge cut a magistrate judge's recommended $7,500 for mitigating circumstances, while warning that fake citations reflect 'professional detachment.'
Influencer Han Anran apologised on May 20 to stars Yang Mi, Yang Zi, Ju Jingyi and Lin Yun after a video on her beauty account showed an AI digital human, speaking as her, rating their 'cosmetic-surgery savvy'. She blamed the avatar's poor wording and said she was on business in France and missed reviewing it.
