A Georgia judge granted OpenAI summary judgment against gun rights activist Mark Walters, who sued after ChatGPT falsely told journalist Frederick Riehl that Walters was accused of embezzlement. Judge Tracie Cason found a reasonable reader in Riehl's position would not take the output as fact, and that Walters showed neither negligence, actual malice nor damages; Riehl never believed it anyway.
China's National Network and Information Security Information Notification Center said on May 20 that Ministry of Public Security testing found 35 apps in the Yingyongbao store illegally collecting and using personal information, including AI assistant Zhipu Qingyan, which collected beyond what users authorised, and Kimi, which collected data with no direct link to its functions.
Who did it: HYBRID AI Hallucination Cases database
Magistrate Judge William Matthewman sanctioned defense attorneys Timothy R. Lord and Joel A. Bello after their April 22, 2025 filing cited a nonexistent case; Lord admitted using AI and not checking the citation. Both must pay Versant Funding's related fees and costs and complete an AI ethics CLE; Lord was fined $1,000 and Bello $500.
Kenyan Supreme Court Justice Isaac Lenaola, chair of the Judiciary's Committee on Innovation, described a case in which lawyers' AI-assisted submissions had 'beautiful' language but every cited authority was fictitious. 'Until we give you guidelines, please avoid AI for now,' Lenaola told judges and lawyers at the Judiciary Digital Transformation Conference.
New York City's Administration for Children's Services uses an algorithm, launched in 2018, that scores families on 279 variables, including neighborhood and mother's age, to flag cases for extra scrutiny, The Markup reports. ACS tells neither families nor their lawyers when it flags a case, and its own audit found likely 'implicit and systemic biases' in the training data.
Dominican Interior and Police Minister Faride Raful released a video denouncing a months-long campaign of discredit and digital blackmail that, Raful says, escalated to using AI to fabricate false rumours about the minister's personal life, and said lawyers have been told to take legal action against everyone involved. 'We know who they are. I'm not afraid of them.'
U.S. District Judge Kathryn Kimball Mizelle struck a motion to dismiss filed for media consultant Tim Burke after finding at least nine nonexistent quotes and misstatements of case law. Lead attorney Mark Rasch, who used ChatGPT's "deep research" and Westlaw's AI features, took sole responsibility. No punishment followed, just a warning: "I expect research to be done by human beings."
Prosecutors say Michael Smith made over $10 million in royalties between 2017 and 2024 by using bots to continuously play AI-generated songs, credited to fake artists like "Calorie Event," on streaming platforms. Arrested on September 4, 2024 in the first US AI streaming fraud case, he pleaded not guilty; Spotify says it paid him about $60,000.
