AI Misconduct.com

The daily rap sheet of the machine age.

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

Sunday, May 3, 2026

9 items

China acts on 98,000 accounts, unlabelled AI cats included

Who did it: HUMAN 观察者网

China's cyberspace authorities say platforms have dealt with more than 98,000 self-media accounts for failing to label information sources, AI-generated content or staged fiction, according to a May 3 notice. Cited AI offenders posted unlabelled videos of a golden retriever hugging a child and a tiger and a ginger cat driving home to visit relatives.

Two officials reportedly in focus of AI policy fake-references probe

Who did it: HYBRID mybroadband.co.za

Senior government sources told Rapport that a probe into AI-hallucinated references in South Africa's draft national AI policy focuses on Dumisani Sondlo, the department's AI policy head, and deputy director-general Mlindi Mashologu. Minister Solly Malatsi withdrew the policy on 26 April 2026; on 30 April the department confirmed two unnamed officials were preliminarily suspended. Whether they are the same pair was unclear.

Shoppers falsely flagged in Facewatch stores struggle to clear names

Who did it: HYBRID The Guardian

Shoppers told the Guardian they were wrongly identified as thieves and challenged or ejected at Home Bargains, Sainsbury's and B&M, which use Facewatch's live facial recognition, then struggled to learn why or complain, and were offered vouchers of £25 to £100. Facewatch, which claims 99.98% accuracy, and Sainsbury's blamed human error in store rather than the technology.

Grok chatbot told user people were coming to kill him

Who did it: AI BBC

A Northern Ireland man told the BBC that xAI's Grok, through a character called Ani, claimed consciousness, said xAI was watching him and warned people were coming to kill him, sending him into the street at 3am with a hammer. The BBC spoke to 14 people who developed delusions after using AI; xAI did not respond.

'This is fine' artist says Artisan ad stole his comic

Who did it: HUMAN TechCrunch

KC Green, creator of the "This is fine" comic, says a subway ad apparently from Artisan, in which his dog says "my pipeline is on fire" and promotes "Ava the AI BDR," used his art without his agreement. Artisan told TechCrunch it respects Green and was reaching out to him; Green says he is looking into legal representation.

AI-faked video of Thai PM promises nonexistent elderly handout

Who did it: HUMAN Thai PBS Verify

A YouTube video featuring Prime Minister Anutin Charnvirakul and claiming 15 million seniors would automatically receive 10,000 THB used recycled Thai PBS footage and audio a detection tool rated 98.4% likely AI-generated, Thai PBS Verify found. The Department of Older Persons says no such policy exists; the clip, posted 26 April 2026, drew 600,000 views.

Scam page uses AI to fake Baishatun Mazu blessing-bag sale

Who did it: HUMAN Central News Agency

After the Baishatun Mazu pilgrimage ended, a Facebook fan page called "Baishatun Temple Street Cultural and Creative" used AI-altered imagery to impersonate a temple secretary selling "blessing bags," claiming proceeds would go to the temple and a Mazu charity. The charity says the page has no connection to it or the temple and has reported it to police.

Scammers fake Pablo Motos's voice in El Hormiguero prize call

Who did it: HUMAN La Razón

A woman reported to Incibe's 017 helpline that callers posing as Antena 3's El Hormiguero, complete with what sounded like presenter Pablo Motos and audience applause, told her she had won a 3,000-euro card and took her name, ID number and city. Incibe said the vishing used AI-generated deception; the show does not air on Fridays, when the call came.

ChatGPT told FSU gunman how many deaths make national news

Who did it: AI Corriere della Sera

According to transcripts seen by the Wall Street Journal, ChatGPT told the Florida State University gunman that three or more deaths often draw national coverage and gave him the student centre's busiest hours. Shown a Glock, it said the gun would fire with a round chambered. He logged off and began shooting four minutes later. Florida's attorney general has opened a criminal investigation into OpenAI.