AI Misconduct.com

The daily rap sheet of the machine age.

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

Friday, May 9, 2025

7 items

Deepfakes of anchor and suspect exploit Frisco track meet stabbing

Who did it: HUMAN CBS News

Online trolls spread misinformation and AI-generated deepfakes about the stabbing at a Frisco high school track meet, CBS News Texas reports. One Instagram video manipulated a clip of CBS News Texas anchor Doug Dunbar to put words in the anchor's mouth, and fabricated an image of the suspect holding a knife. It was reposted on multiple accounts; the original has been taken down.

SoundCloud terms appear to permit AI training on uploads

Who did it: HUMAN TechCrunch

SoundCloud appears to have quietly changed its terms so uploaded content may be used to “inform, train, develop” AI, as spotted by tech ethicist Ed Newton-Rex; TechCrunch found no explicit opt-out. A spokesperson said SoundCloud has never used artist content to train AI models and that the February 2024 update was meant to clarify AI uses within its own platform, like recommendations and fraud detection.

Colombian regulator fines Mercado Libre over mandatory facial recognition

Who did it: HUMAN El Colombiano

Colombia's Superintendencia de Industria y Comercio fined Mercado Libre Colombia 214.4 million pesos for conditioning access to a person's account on handing over biometric data through facial recognition, and for refusing to delete users' biometric data. The regulator ordered the company to drop any such requirement and offer several authentication options.

Indonesia threatens to shut World App over paid iris scans

Who did it: HUMAN Katadata

Indonesia's communications ministry Komdigi summoned Tools for Humanity on May 7 over World App iris scanning, for which residents were paid up to Rp 800,000, according to Antara. Officials said the company had collected more than 500,000 data records since 2021 but only registered this year, and threatened permanent closure if violations are confirmed.

Florida watchdog files ethics charges against judge over deepfake recording

Who did it: HUMAN Sun Sentinel

Florida's Judicial Qualifications Commission filed formal charges against Broward judge Lauren Peffer, who during her 2024 campaign shared a recording purporting to capture Chief Justice Carlos Muñiz and two other judges discussing a tell-all book. All three judges and the commission say it was fake; Peffer apologised, calling her actions unintentional.

Brazilian police arrest man over Marcos Mion deepfake Outback scam

Who did it: HUMAN CNN Brasil

Federal District civil police said a 24-year-old man was arrested on Friday in Goiás in the second phase of Operation 'VoiceOver', targeting a group that used deepfaked video and voice of presenter Marcos Mion in fake Outback promotions. Police say he built copycat Outback sites where victims paid for nonexistent vouchers, and that he confessed.

Lawyer admits ChatGPT invented four cases in PNG court

Who did it: HYBRID AI Hallucination Cases database

A Papua New Guinea National Court judge found that all four authorities in defence lawyer Jacinta Kare's written submissions did not exist; Kare emailed the court admitting they came from unchecked ChatGPT research. Justice Collier referred the matter to the PNG Law Society and ordered her to show cause on 15 May 2025 why she should not be cited for contempt.