AI Misconduct.com

The daily rap sheet of the machine age.

Updated Nov 2, 2026
4,173 items in the archive
RSS feed

Friday, March 27, 2026

7 items

Gemma O'Doherty's AI-written submissions cited nonexistent cases, court says

Who did it: HYBRID The Irish Times

The Court of Appeal dismissed self-represented journalist Gemma O'Doherty's appeal in a defamation case brought by councillor Jimmy Guerin, noting her AI-prepared submissions cited authorities that "simply did not exist". Judge Caroline Costello said she did not believe O'Doherty intended to mislead, and set out principles urging litigants to disclose AI use and verify their submissions.

Polish data watchdog refers AI nude of teacher to prosecutors

Who did it: HUMAN RMF24

Poland's data protection authority UODO notified the Warsaw-Śródmieście district prosecutor after a teacher's photo was "undressed" with AI and posted online. UODO president Mirosław Wróblewski argued this was unlawful processing of personal data under article 107 of the data protection act, punishable by a fine, restriction of liberty or up to two years' imprisonment.

Insurer sues OpenAI, claiming ChatGPT practiced law without a license

Who did it: HYBRID The Indiana Lawyer

Nippon Life Insurance Company of America has sued OpenAI in federal court in Illinois, alleging ChatGPT practiced law without a license by helping a policyholder who had settled with the insurer try to reopen the case; the suit says she filed nearly 50 motions created with ChatGPT. It also claims tortious interference and abuse of process.

Third Circuit reprimands lawyer who cited AI-invented DEA ruling

Who did it: HYBRID AI Hallucination Cases database

The Third Circuit reprimanded attorney Daniel A. Pallen, whose brief in McCarthy v. DEA relied on AI-generated summaries from a non-attorney: seven were inaccurate and one cited adjudication did not exist. He called the errors "immaterial" without checking them. The court said monetary sanctions would likely have followed but for mitigating factors, including this being its first opportunity to address AI.

NJ appeals court sanctions lawyer $1000 over four phantom cases

Who did it: HYBRID AI Hallucination Cases database

The New Jersey Appellate Division sanctioned plaintiff's counsel in Amtrust v. Liberty Mutual $1000 for citing four non-existent cases, which it called "hallmarks of . . . generative artificial intelligence." Told one case could not be found, counsel replied that legal databases confirmed "a case . . . with a similar citation"; the court said it was "stunned by the carelessness."

Charities condemn AI-made fetishized images of disabled women

Who did it: HUMAN Le Monde

British charities condemned a trend of sexualized AI images depicting women with Down syndrome, vitiligo and albinism, often unlabeled and drawing millions of views; some accounts alter real images of non-disabled women. The Down's Syndrome Association called it "a scam", and Higgsfield, a virtual-model platform, offers options to add vitiligo and albinism.

iyO amends OpenAI lawsuit to allege trade secret theft

Who did it: HUMAN 9to5Mac

In an amended complaint filed March 13, hearing device startup iyO alleges a former iyO engineer downloaded confidential CAD files for io co-founder Tang Tan, now a defendant, who allegedly reviewed internal materials and prototypes at a June 2024 dinner. iyO claims the misappropriation by Tan and OpenAI accelerated io Products' development 'by nearly a decade'.