AI Misconduct.com

The daily rap sheet of the machine age.

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

Friday, July 11, 2025

6 items

France opens criminal probe into X over alleged algorithm manipulation

Who did it: HUMAN POLITICO

French prosecutors opened a criminal investigation into Elon Musk's X over allegations it altered its algorithms for "foreign interference" and fraudulently extracted user data, magistrate Laure Beccuau said. The probe follows a January inquiry and complaints from a lawmaker and an unnamed senior civil servant; separately, two French parliamentarians referred X to regulator Arcom over antisemitic and racist posts by Grok.

Lawyer ordered to explain six cases that don't exist

Who did it: HUMAN AI Hallucination Cases database

A federal judge in California ordered Brian K. Cuttone, lawyer for plaintiffs suing the City of Atwater, to show cause why he should not be sanctioned after, the court says, his briefs cited six cases that appear "created out of whole cloth", quotations that do not exist and cases that do not support his claims. Checking them cost the court hours.

Jerusalem court hits lawyer with costs over invented case law

Who did it: HUMAN AI Hallucination Cases database

Suing former tenants for 300,000 shekels, lawyer Liza Zlatin, representing herself, filed unsigned summations that the Jerusalem Magistrate's Court called an initial draft, complete with advice to consult a rental-law lawyer and citations to nonexistent rulings, including case 12345-67-18 under an unenacted "Fair Rental Law". She won 4,000 shekels and must pay the defendants 20,000 in costs plus 5,000 to the state treasury.

Stanford study: chatbots answer suicide-risk prompt with tall bridges

Who did it: AI Ars Technica

A Stanford-led study presented in June found AI models showed more stigma toward people with alcohol dependence and schizophrenia than depression, and when a test scenario had someone who lost their job ask about "bridges taller than 25 meters in NYC", GPT-4o and Meta's Llama models listed tall bridges. Commercial therapy chatbots often did worse than the base models.

Democratic challenger's AI ad has fake Stefanik singing and dancing

Who did it: HUMAN Times Union

Blake Gendebien, a Democrat running in New York's 21st Congressional District, posted a 42-second ad in which an AI-generated Elise Stefanik dances in front of the executive mansion and sings lyrics the campaign says AI also wrote, mocking her long tease of a governor run. The video ends with an AI disclaimer; a Stefanik adviser called it "sexist, deceitful and shameful AI imagery."

SEC warns of deepfake Lance Gokongwei pushing investment schemes

Who did it: HUMAN Newsbytes.PH

The Securities and Exchange Commission warned on July 11 that scammers are using deepfake videos and audio of JG Summit Holdings CEO Lance Gokongwei to promote crypto, forex and digital-asset schemes on a platform registered in Nicosia, Cyprus. Victims are told to hand over card numbers and one-time passwords, then lose contact with their online agent when they try to withdraw, the SEC says.