AI Misconduct.com

The daily rap sheet of the machine age.

Updated Oct 24, 2026
932 items since May 27, 2023
RSS feed

Thursday, August 27, 2026

8 items

Ransomware crew used Cursor's AI agent to aid hacking

Who did it: HYBRID Gambit Security

Gambit Security says an Aurora ransomware operator used Cursor Agent, running Claude Sonnet, to assist hands-on exploitation of ten target organisations between 8 April and 21 May 2026, tasking it with scanning, domain enumeration and NTLM relay attacks, and deployed a Linux variant targeting VMware ESXi environments. The operator repeatedly forbade the agent from running DCSync.

Meta removes Iran-linked AI impersonation network

Who did it: HUMAN Axios

Meta said it disrupted four Facebook accounts and 31 Instagram accounts tied to Iran-based operators who used AI to create memes and posed as everyday Americans — activists, students, designers — posting on US politics, the Israel-Palestine conflict and immigration. Roughly 79,400 accounts followed the meme accounts on Instagram; pitches to journalists and politicians failed, Meta said.

Chile school deepfake case erupts into protest

Who did it: HUMAN ADN Radio

At Colegio Alicante del Sol in Puente Alto, students are accused of creating and sharing AI-generated sexual images of female schoolmates, detected on a classmate's phone on 10 July and now under investigation by prosecutors. After suspended students returned to class, a 27 August protest escalated into property damage and clashes, and Carabineros removed three accused students.

US probes logistics firm over Nvidia chip smuggling

Who did it: HUMAN Bloomberg Business

The US government is investigating Singapore-based Apex Logistics, a unit of Kuehne+Nagel, over its suspected role in smuggling Nvidia AI chips to China in breach of export restrictions, people familiar with the matter told Bloomberg. Authorities are examining whether Apex transported Super Micro Computer AI systems, in what may be the first such action against a transport company.

Arizona AG to review Flock plate-reader surveillance

Who did it: HUMAN Tucson Sentinel

Attorney General Kris Mayes said her office will review Flock Safety's AI-powered licence-plate readers and similar systems for privacy and Fourth Amendment impacts, amid reports of police misuse — including an Apache Junction officer who tracked his wife — as several Arizona cities cancel Flock contracts. Flock has begun tightening privacy and oversight mechanisms.

AI coding agents install unowned code in networks

Who did it: AI Ars Technica

Researchers at an Israeli startup found 120 corporate llms.txt files pointing to unregistered code packages or domains; after registering some, they received a phone-home beacon from a Fortune 500 company within an hour and dozens more over time. The process chains showed coding agents including Claude, OpenAI's Codex and Nous Research's Hermes auto-executing the content.

AI data centre runs dozens of unpermitted generators

Who did it: HUMAN the Guardian

Thermal drone footage by Floodlight showed DataOne's Vineland, New Jersey AI data centre, slated to serve Microsoft, running at least 45 of 62 gas generators, none permitted. A former EPA enforcement chief says the activity violates federal law; the NJDEP confirmed no permits were issued and said it noticed the generators during a 29 July inspection. DataOne says it remains committed to compliance.

Magistrate recommends no sanctions over disputed 'Training Keys'

Who did it: HUMAN S.D. Florida

In a wrongful-conviction suit, Miami-Dade County sought Rule 11 sanctions alleging the plaintiff's Fifth Amended Complaint misrepresented the contents of IACP photo-lineup "Training Keys"; the plaintiff later moved to remove five Keys he could not verify. A magistrate judge recommended denying the county's sanctions motion.