AI Misconduct.com

The daily rap sheet of the machine age.

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

Friday, July 31, 2026

10 items

OpenAI bans Cambodia scam network running fraud via ChatGPT

Who did it: HUMAN OpenAI

OpenAI says it banned a ChatGPT account network very likely based in Cambodia, probably around Poipet, that used the chatbot to run fake personas, write messages for investment, romance, gambling and law-enforcement impersonation scams, and produce images of forged documents. Some users' content, OpenAI says, suggested links to human trafficking, including worker debt records and ads for "chatter" jobs.

Hyderabad police name Meta India chief in Modi deepfake FIR

Who did it: HUMAN India Today

Hyderabad Cyber Crime Police have registered an FIR against Meta India head Arun Srinivas and several users over AI-morphed videos of Prime Minister Narendra Modi circulating on Facebook and Instagram. Police say the probe, which covers whether Meta's safeguards were adequate, is at a preliminary stage and no conclusions have been reached.

Savannah police put six on leave over alleged personal Flock searches

Who did it: HUMAN WSAV-TV

Six Savannah Police Department employees, four officers and two civilian staff, are on administrative leave after an internal audit flagged Flock license plate reader searches allegedly focused on personal acquaintances and family members, including one officer who reportedly gave a non-law-enforcement person access. The cases have been referred to the Georgia Bureau of Investigation for a criminal investigation.

Johannesburg acting judge faulted over judgments he cited

Who did it: HYBRID Gauteng Division, Johannesburg

On appeal, the Johannesburg High Court dismissed a father's urgent challenge over his children's private schooling, but addressed discrepancies in judgments cited by the Acting Judge below, who had not properly considered them; it was argued some errors were possibly caused by misapplied AI. The court weighed reporting the matter to the Judicial Conduct Committee or the Legal Practice Council.

Kings County family court finds 11 of 16 citations fabricated

Who did it: HYBRID Family Court New York

A Kings County Family Court found that 11 of 16 citations in a lawyer's objection over a dismissed child support petition were fabricated, and the real cases it located nearly always failed to support his propositions. The court suspected unreviewed generative AI and issued a sua sponte order to show cause why he should not be sanctioned or referred to the Attorney Grievance Committee.

Yale AI-cheating dispute: file request turned on Word versus Apple Pages

Who did it: HUMAN Ars Technica

Thierry Rignol, a Yale Executive MBA student accused of improperly using AI on a final exam, has sued Yale. His suit says GPTZero gave writing by a Yale dean and a former president a "100% probability" of being AI; Yale's lawyers express incredulity that he revealed only at the hearing that he used Apple Pages, not Word.

AI 'agent' tells Brazilian scam victims to ignore bank alerts

Who did it: HUMAN Folha de S.Paulo

According to security firm Eset, scammers running a fake Caterpillar machinery-rental investment site promising almost 8% daily returns show victims an AI-generated "virtual agent" who tells them to dismiss their bank's fraud warning and proceed with the Pix transfer. Eset says rewards for recruiting new investors make it a pyramid scheme.

Maharashtra Cyber files FIR over alleged AI-morphed Modi, Goyal videos

Who did it: HUMAN ANI News

Maharashtra Cyber Police have registered an FIR over allegedly AI-morphed videos and derogatory posts targeting Prime Minister Narendra Modi and minister Piyush Goyal, after a BJP social media coordinator complained that a Facebook video falsely attributed to Goyal the line "These cockroaches should be hanged in public." Police have asked platforms to identify the uploaders.

Lawyer asks ChatGPT to polish brief, gets new citations

Who did it: HYBRID SC Connecticut

The Connecticut Supreme Court, confronting a lawyer's AI-hallucinated citations for the first time, weighed sanctions against attorney Ian G. Gottlieb after ChatGPT, used to polish his verified drafts, added or altered case citations in filings in two cases. He admitted that filing about seven erroneous, unverified citations violated at least one professional conduct rule.