AI Misconduct.com

The daily rap sheet of the machine age.

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

Wednesday, December 24, 2025

9 items

Italy orders Meta to suspend WhatsApp terms excluding rival chatbots

Who did it: HUMAN Reuters

Italy's antitrust authority AGCM ordered Meta Platforms on Dec 24 to suspend WhatsApp contractual terms that, the watchdog said, "completely exclude Meta AI's competitors" from the platform, as it investigates suspected abuse of a dominant position. A Meta spokesperson called the decision "fundamentally flawed" and said: "We will appeal."

Korean student booked over AI deepfake nudes of teachers

Who did it: HUMAN The Chosun Ilbo

South Korean police have booked, without detention, a second-year high school student suspected of using AI deepfake technology from 2022 to this year to put the faces of 10 people, including two teachers and students, onto nude images. Police say forensic analysis of the student's phone found more than 100 such images, and further investigation, including into distribution, is planned.

Defendant gets two years for ChatGPT-doctored insurance paperwork

Who did it: HUMAN Yonhap News Agency

The Busan District Court sentenced a person in their 20s to two years in prison for insurance fraud. The defendant uploaded photos of hospital admission records to ChatGPT and asked it to lengthen the stays, collecting 150 million won over 11 claims from July 2024, including one claiming an acquaintance was hurt playing football. The judge noted no settlement with the insurers.

Lawsuit says Fireflies.AI harvests meeting voiceprints without consent

Who did it: HUMAN Data Privacy + Cybersecurity Insider

A proposed class action in Illinois federal court alleges Fireflies.AI Corp.'s meeting assistant collects and stores participants' biometric voice data for speaker recognition without consent, and publishes no retention or destruction policy as Illinois' BIPA requires. The plaintiff, an Illinois resident, says her voice was profiled in a November meeting though she never used the service.

Beeping Bluetooth gives away AI cheating in Hyderabad job exam

Who did it: HUMAN Telangana Today

Gachibowli police arrested two candidates from Haryana accused of cheating in a University of Hyderabad junior office assistant recruitment exam. Police say a miniature scanner on a shirt button captured the questions and AI-generated answers were relayed to a Bluetooth device; an invigilator caught the first after hearing repeated beeping.

Deepfaked MP 'endorses' eyedrops he says don't work

Who did it: HUMAN Balkan Insight

Surgeon and MP Vladimir Dobricanin found an AI-manipulated video on fake Facebook and Instagram profiles showing him endorsing eyedrops and remedies for rheumatism, impotence and parasites, with ads linking to a site of Limited Charm, registered as Digital Edge. Montenegro's CIRT says misuse of public figures' likenesses has grown since 2022.

Fair Work Commission sets aside AI-assisted case citations

Who did it: HYBRID AI Hallucination Cases database

In an unfair dismissal claim against GTC Contracting, Fair Work Commission Deputy President Butler suspected the applicant's husband and representative, Michael Archer, had used AI for his closing submissions, which he confirmed. The Commission disregarded cited cases for which he could not find citations or links, and dismissed the application on 24 December 2025.

Tribunal finds EI appellant's four cited cases do not exist

Who did it: HYBRID AI Hallucination Cases database

Canada's Social Security Tribunal found that four Federal Court of Appeal cases cited by an employment insurance appellant, including "Gagnon v Canada, 1999 FCA 8976", do not exist, their citations leading to unrelated decisions. Member Gary Conrad disregarded them and on December 24, 2025 dismissed the appeal seeking to backdate the claim.

LivCor reaches proposed DOJ settlement in RealPage rent case

Who did it: HUMAN PYMNTS.com

LivCor, one of six landlords the Justice Department and states sued in January for allegedly using rivals' sensitive information through RealPage pricing algorithms to set rents, reached a proposed settlement barring anticompetitive algorithms and requiring a court-appointed monitor if it uses an uncertified pricing algorithm. LivCor notes the deal contains "no finding of or admission of any wrongdoing."