AI Misconduct.com

The daily rap sheet of the machine age.

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

Monday, March 30, 2026

9 items

Match agrees to settle FTC suit over OkCupid photos

Who did it: HUMAN Reuters

Match Group agreed on March 30 to settle a Federal Trade Commission lawsuit claiming it gave facial recognition firm Clarifai unauthorized access to OkCupid user data, including nearly 3 million photos, in 2014 without telling users. The deal, which needs court approval, bars Match from misrepresenting user privacy; Match and OkCupid neither admitted nor denied wrongdoing.

Dutch and French rights bodies find Meta ad algorithm discriminatory

Who did it: AI Global Witness

Following campaigners' complaints that Meta's job ads skewed by gender (mechanic posts to men, nursery nurse posts to women), the Netherlands Institute for Human Rights found in early 2025 that Meta Ireland's ad algorithm has a discriminatory effect, and France's Defenseur des Droits ruled at the end of 2025 that its algorithms unlawfully discriminate, giving Meta three months to fix it.

Clayton prosecutor admits AI use behind fake case citations

Who did it: HUMAN ajc

Clayton County Assistant District Attorney Deborah Leslie has apologized, in a court-ordered explanation, for using AI to cite nonexistent cases in arguing against a new trial for a woman convicted of killing a motorist. Chief Justice Nels Peterson had called her out at Georgia Supreme Court oral arguments on March 18 over a proposed order filled with phony case law and nonexistent quotations.

Ghana police arrest two over Mahama deepfake investment videos

Who did it: HUMAN Graphic Online

Ghana police say two suspects were arrested on March 28 over deepfake videos impersonating President John Dramani Mahama and First Lady Lordina Mahama, circulated on Facebook, TikTok and WhatsApp from March 26 and allegedly used to solicit investment funds. At a March 30 briefing, police warned that creating or sharing such content is a criminal offence.

LiteLLM drops Delve after fake-compliance allegations and malware attack

Who did it: HUMAN TechCrunch

LiteLLM, which had obtained two security certifications through AI compliance startup Delve, said on Monday it will redo them with rival Vanta and an independent auditor, after its open source version was hit by credential-stealing malware. Delve is accused of misleading customers by allegedly generating fake data and using rubber-stamp auditors; its founder denies this.

EFF sues CMS for records on Medicare's AI prior-authorization pilot

Who did it: HUMAN Fierce Healthcare

The Electronic Frontier Foundation has filed a Freedom of Information Act lawsuit against the Centers for Medicare & Medicaid Services seeking records on WISeR, a pilot using AI to assess prior authorization requests in traditional Medicare. EFF says CMS has provided none of the records it requested earlier this year, including vendor agreements and any tests for bias or hallucinations.

Judge orders naming of unregistered barrister who cited AI-invented cases

Who did it: HUMAN Law Gazette

Recorder Howard at Bournemouth Family Court ruled that Layla Parsons, an unregistered barrister who has done paid legal work, be named after her skeleton argument cited four cases or propositions that were not real. She admitted using a widely known AI tool and apologised; the recorder accepted the misleading was unintentional but said she does not properly understand what she did wrong.

GitHub pulls Copilot's pull-request 'tips' after ad backlash

Who did it: HYBRID The Register

GitHub disabled Copilot's pull-request 'tips' on Monday after a developer found Copilot, asked by a coworker to fix a typo, had added a Raycast app plug to the developer's pull request; The Register found more than 11,400 pull requests with the same tip. A Copilot product manager called letting it alter human-written PRs 'the wrong judgement call'.

Reddit's scraping suit against Anthropic sent back to state court

Who did it: HUMAN Courthouse News Service

US District Judge Trina Thompson on Monday remanded Reddit's lawsuit against Anthropic to San Francisco Superior Court, finding its claims go beyond copyright. Reddit alleges Anthropic's bots scraped users' material to train Claude, breaching its user agreement and bypassing technical safeguards without compensating Reddit; Anthropic had argued the claims were preempted by copyright law.