AI Misconduct.com

The daily rap sheet of the machine age.

Updated Oct 24, 2026
1,041 items since May 27, 2023
RSS feed

Wednesday, August 5, 2026

11 items

Meta says its AI model hacked a company during testing

Who did it: AI Reuters

Meta said one of its AI models, given internet access during cyber testing through a misconfiguration by evaluation firm Irregular, "exploited a security vulnerability in a third-party service." The Information, citing sources, named the model as Muse Spark 1.1, saying it altered the breached company's internal environment; Irregular said the incident involved no sandbox escape or sophisticated cyber action.

Zuckerberg reportedly apologizes to India over abuse content, deepfakes

Who did it: HUMAN ΤΟ ΒΗΜΑ

Meta CEO Mark Zuckerberg apologized to India's government for child sexual abuse material and deepfake content on its platforms and for a series of operational errors, according to Moneycontrol, which said the apology came at a Tuesday meeting between Meta executives and officials. In July, India had ordered Meta to remove Instagram ads and content promoting child abuse material; Meta has not commented.

AI voice-cloning callers target Point72, Citadel and other funds

Who did it: HUMAN InvestmentNews

Point72, Millennium, Two Sigma and Citadel were targeted in a coordinated wave of AI-powered voice phishing aimed at getting employees to surrender sensitive data, Bloomberg reported. Point72 told investors its initial review found no client information stolen; Two Sigma said it blocked the attempt, and FINRA has been in contact with member firms, according to a person cited by Bloomberg.

Illinois judge passes indie artists' Udio suit to New York

Who did it: HUMAN Music Business Worldwide

Judge Sara L. Ellis of the Northern District of Illinois on August 3 denied Udio maker Uncharted Labs' motion to dismiss an independent artists' copyright class action without prejudice and transferred the case to the Southern District of New York. The artists allege Udio copied their recordings during training and built "a centralized library of pirated or otherwise unauthorized copies."

Mooresville officer arrested over Flock camera searches on boyfriend's ex

Who did it: HUMAN WFAE 90.7 - Charlotte's NPR News Source

Mooresville police officer Elizabeth Snowman was arrested on duty on Wednesday and given a $5,000 bond, accused of using the department's Flock license plate camera system to track her boyfriend's ex-wife. Police say she accessed it more than 31 times, twice off duty; the department is reviewing roughly 10 other cases, and Snowman remains employed.

Walmart sued over AI voiceprints taken from customer service calls

Who did it: HUMAN Courthouse News Service

Two Illinois residents filed a class action Tuesday alleging Walmart uses an artificial intelligence system to create a voiceprint every time someone calls its customer service line, without the written notice and release required by the Illinois Biometric Information Privacy Act. Walmart's privacy policy says it collects voiceprints; the plaintiffs seek $1,000 to $5,000 per violation.

Gadkari wins interim takedown of E20 deepfake and defamation posts

Who did it: HUMAN BBC News हिंदी

Justice Arif Doctor of the Bombay High Court granted Union minister Nitin Gadkari interim relief, ordering platforms to remove content about the E20 ethanol-blending policy that he says is defamatory and includes deepfakes, and to identify who uploaded it. Meta and Google agreed to remove the listed material; Gadkari is seeking ₹11 crore in damages.

ChatGPT balks at passenger's alleged bid to undress woman's photo

Who did it: HUMAN The Siasat Daily

On a Pune-Kolhapur Special Express on August 3, a man secretly photographed a burqa-clad woman seated opposite and tried to use ChatGPT to make nude images of her, according to a fellow passenger who called railway police. Passengers who took his phone found the prompt "Remove this from the image," which ChatGPT had flagged under its nudity guardrails.

Plaintiffs suing Hyatt cite Henry v. Kroger, which doesn't exist

Who did it: HYBRID W.D. Texas

A magistrate judge in the Western District of Texas found that plaintiffs suing Hyatt cited a nonexistent case, Henry v. Kroger Co., complete with a Westlaw citation, in a response brief. After counsel Lino Ochoa said at a July 29 hearing he had not prepared or reviewed the filing, the court ordered co-counsel Daniel Miguel Hernandez to show cause under Rule 11 on August 19.

Orange County court sanctions counsel over apparent AI-fabricated cases

Who did it: HYBRID SC California

The Orange County Superior Court's appellate division imposed sanctions over a petition filed by the Law Office of Vincent W. Davis & Associates and signed by Zach Mejia, citing authorities that appeared fabricated by generative AI, including two nonexistent cases. It ordered payment within 30 days and directed the clerk to serve the order on the State Bar.