AI Misconduct.com

The daily rap sheet of the machine age.

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

Monday, December 8, 2025

7 items

Pakistan court reinstates 36 teachers after tribunal cited phantom cases

Who did it: HYBRID The Express Tribune

Pakistan's Supreme Court ordered 36 secondary school teachers in Khyber-Pakhtunkhwa reinstated, overturning a July 14, 2025 Service Tribunal judgment that upheld their termination. The court noted with "dismay" that the tribunal relied on legal citations that do not exist, and suggested it may have used artificial intelligence that "hallucinated" them; it also found the education directorate had violated due process.

Privacy commissioner probes Toronto billboards that guess passersby's age

Who did it: HUMAN CP24

Canada's privacy commissioner, Philippe Dufresne, has opened an investigation, following complaints, into digital signs near Toronto's Union Station that allegedly use facial detection software to count audiences and estimate gender and approximate age. The signs were installed about three years ago by Cineplex Digital Media, which has said it strictly adheres to the commissioner's guidance.

Bunnings chatbot walks customer through electrician-only plug job

Who did it: AI Inside Retail Australia

Bunnings has admitted its AI chatbot gave faulty advice after it gave a Queensland customer six steps for replacing an extension cord plug, work that only a licensed electrician may legally do there. Chief information officer Gen Elliott said the company has strengthened safeguards so requests about electrical, plumbing or other licensed work are referred to a qualified tradesperson.

Alberta court adds costs for litigant's fake legal authorities

Who did it: HUMAN AI Hallucination Cases database

Alberta's Court of King's Bench added $1,000 per set of respondents to the costs owed by self-represented appellant Joel Barrette, who cited at least three fake authorities: two non-existent cases and a non-existent regulation. Applying a Court of Appeal precedent on AI-generated citations, Justice Douglas R. Mah noted Barrette never acknowledged sending three sets of counsel and the court on a wild goose chase.

Arizona lawyer checked fake citations by asking ChatGPT

Who did it: HYBRID AI Hallucination Cases database

Arizona attorney Matthew V. Moosbrugger admitted at a hearing that he checked a brief's AI-generated citations, from Federally Lawyer, a ChatGPT add-on, only by asking ChatGPT whether they were real cases; the court found one non-existent case and five misquoted ones. Deeming public humiliation a sufficient remedy, it imposed no further sanctions but sent its order to the State Bar of Arizona.

UK's Cooper to warn of foreign AI videos undermining Ukraine

Who did it: HUMAN The Guardian

UK Foreign Secretary Yvette Cooper will warn on Tuesday that foreign states are flooding social media with generative AI and manipulated videos to undermine support for Ukraine, in remarks apparently aimed at Moscow. Officials warn Russia is using forged documents and deepfake material, and point to fake websites mimicking Moldova's PAS party during September elections.

Deepfake videos falsely show MC Quyền Linh being arrested

Who did it: HUMAN VietNamNet News

From the afternoon of December 7, fanpages shared videos headlined 'Police urgently arrest MC Quyền Linh for selling poor-quality goods', showing a man resembling the host being restrained; users soon spotted them as deepfakes. In a December 8 livestream, Quyền Linh called the claims baseless, saying he had work engagements in Hanoi with the Vietnam Cinema Association.