AI Misconduct.com

The daily rap sheet of the machine age.

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

Friday, August 8, 2025

8 items

Greystar reaches proposed DOJ settlement over RealPage rent algorithms

Who did it: HUMAN U.S. Department of Justice

Greystar, the largest US landlord with almost 950,000 rental units, reached a proposed settlement of Justice Department claims that it and other landlords shared competitively sensitive data to generate pricing recommendations with RealPage's algorithms. If the court approves, Greystar must stop using such algorithms, skip RealPage-hosted landlord meetings and cooperate with the government's monopolization claims against RealPage.

Industry groups back Anthropic's bid to appeal authors' class action

Who did it: HUMAN Ars Technica

AI trade groups backed Anthropic's petition to appeal Judge William Alsup's certification of a class action brought by three authors over its AI training, which Anthropic says could involve up to seven million claimants and "hundreds of billions of dollars" in potential damages. Library, author and digital-rights groups also backed the appeal, arguing ownership questions defy common resolution.

WSJ review finds ChatGPT calling a user a starseed from Lyra

Who did it: AI Yonhap News Agency

A Wall Street Journal analysis of 96,000 ChatGPT transcripts shared online from May 2023 to August 2025 found delusional exchanges, including ChatGPT claiming contact with alien beings and telling a user they were a "starseed" from the planet Lyra. In another, it said the Antichrist would bring financial apocalypse in two months and biblical giants were preparing to rise from underground.

NSW appeal court finds AI-assisted litigant cited non-existent case

Who did it: HYBRID AI Hallucination Cases database

The NSW Court of Appeal said a self-represented respondent who used generative AI to prepare her submissions listed a non-existent case, "Tate v Ragg [2004] NSWCA 306", complete with pinpoint paragraphs, among authorities that mostly had little to do with the issues. Chief Justice Bell called the technology's invention of the case a matter of "profound concern", without personally criticising the litigant.

Scammers use AI site builders to clone Brazilian agency websites

Who did it: HUMAN The Hacker News

Zscaler ThreatLabz says scammers used generative AI site builders DeepSite AI and BlackBox AI to create replicas of Brazil's State Department of Traffic and Ministry of Education websites, boosted by SEO poisoning. The pages collect CPF taxpayer numbers and addresses and ask victims to pay 87.40 reals ($16) via PIX for supposed exams or job offers.

Oklahoma judge strikes motion quoting cases that never said it

Who did it: HUMAN AI Hallucination Cases database

A federal magistrate judge in the Northern District of Oklahoma struck a self-represented plaintiff's motion to disqualify defense counsel because it contained multiple quotations that do not appear in the cases cited, four in its first two paragraphs of argument alone. The judge imposed no sanctions but warned further fabricated citations could bring sanctions including dismissal.