AI Misconduct.com

The daily rap sheet of the machine age.

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

Wednesday, August 6, 2025

9 items

Report: Israel's Unit 8200 stores Palestinian calls in Microsoft cloud

Who did it: HUMAN The Guardian

A Guardian investigation with +972 Magazine and Local Call reports that Israel's Unit 8200 has used a segregated area of Microsoft's Azure cloud since 2022 to store recordings of millions of Palestinian phone calls made each day. Three unit sources say it helped prepare deadly airstrikes. Microsoft says it had "no information" about the data stored and was unaware of civilian surveillance.

Meta AI contractors say they read users' identifiable intimate chats

Who did it: HUMAN Business Insider

Four contractors hired through Outlier, owned by Scale AI, and Alignerr told Business Insider they routinely saw chats between users and Meta's AI containing full names, phone numbers and email addresses; two said users sent selfies. One estimated personal data appeared in more than half of the chats. Meta says it "intentionally limit[s]" what personal information contractors see.

Documents tie Chinese AI firm GoLaxy to influence campaigns

Who did it: HUMAN The New York Times

Internal GoLaxy documents obtained by Vanderbilt University researchers describe a 'Smart Propaganda System' and influence campaigns in Hong Kong and Taiwan. They also claim the company built profiles on 117 current and former members of Congress, according to the New York Times. GoLaxy denies doing any work related to Hong Kong or elections and called the Times' information 'misinformation'.

Spain fines Supera gym chain over mandatory facial recognition entry

Who did it: HUMAN elDiario.es

Spain's data protection agency AEPD fined Sidecu, operator of the Supera gym chain, €160,000 for making facial recognition initially the only way into several gyms from summer 2023. The agency found the biometric templates were special-category data. The penalty fell to €96,000 after the company paid voluntarily and acknowledged responsibility. The AEPD ordered the system's immediate temporary suspension.

Brazil keeps ban on World paying people for iris scans

Who did it: HUMAN Folha de S.Paulo

Brazil's data protection authority ANPD rejected an appeal by Tools for Humanity and the World Foundation. Its ban on paying Brazilians in cryptocurrency for biometric data, including iris scans, stays in place; the ANPD says payment can undermine free consent. Tools for Humanity says it respectfully disagrees and plans further legal measures. More than 400,000 Brazilians had registered.

Lawmaker alleges Indiana lieutenant governor's aide made deepfake of wife

Who did it: HUMAN Fox 59

An Indiana state representative says a staffer in Lt. Gov. Micah Beckwith's office allegedly created a pornographic deepfake video of the lawmaker's wife and showed it to other staffers. The lawmaker has not seen the video. Marion County Prosecutor Ryan Mears would neither confirm nor deny an investigation. According to the lawmaker, Beckwith doesn't believe the allegations are true.

Lawyer in T.D. Jakes defamation case sanctioned over AI hallucinations

Who did it: HYBRID MinistryWatch

U.S. District Judge William Stickman ordered New York attorney Tyrone Blackburn, who defended the man T.D. Jakes is suing for defamation, to pay over $76,000 in Jakes's fees. The fees covered responding to 'AI-generated hallucinations and other misrepresentations' in a motion to dismiss. Blackburn said he bought an AI feature because he employs no paralegals.

Haifa court: 31 of 32 cited judgments don't exist

Who did it: HYBRID AI Hallucination Cases database

In a suit against the Israel Land Authority, Haifa District Court Judge Esperanza Alon found that 31 of 32 judgments and quotations cited by the plaintiff did not exist and appeared to have been invented by an unreliable computerised system. The judge rejected the lawyer's reliance on a secretary. The judge also said such conduct justifies dismissing the suit with costs.