The Hall of Shame

The legal complaint filed against Koi Security (now owned by Palo Alto Networks), its researchers, and Palo Alto Networks alleges that Koi used an LLM to generate the threat report, the AI system hallucinated findings about MeetingTV, and the security shop then published those as facts in a December 30 blog. Accuses Koi (now Palo Alto Networks) of “reckless publication of an AI-driven cybersecurity report that falsely accused Plaintiff MeetingTV Inc. of criminal conduct including operating core infrastructure for a well-funded Chinese criminal organization running a large-scale malware and corporate espionage campaign, being a "front" for a chinese cybercriminals.

Starbuck alleges Google's AI programs including Gemini hallucinated false criminal allegations including sexual assault, child rape, abuse, fraud, stalking, drug charges, and even saying he was in Epstein’s flight logs.

Yale suspended a top Executive MBA student for a year and gave him a failing grade after the AI-detection tool GPTZero flagged his final exam as likely AI-generated. The student, a non-native English speaker whose formal prose GPTZero is known to misclassify, disputes the finding and has filed a 13-count federal lawsuit against the university alleging breach of contract, civil rights violations, defamation, and invasion of privacy.

Its website chatbot incorrectly told a customer he could claim a bereavement fare discount retroactively after travel (contradicting the actual policy). A Canadian tribunal held the airline liable and ordered compensation.

A cybersecurity/loyalty programs report was withdrawn after researchers found extensive fabricated citations, nonexistent sources, and inaccurate data (GPTZero analysis alleged ~60% hallucinated references in one report).

An agentic AI / customer experience report was retracted after claims about companies (e.g., UBS, NHS) and the majority of citations were flagged as fabricated or inaccurate.

A government-commissioned report contained fabricated academic citations and other errors linked to generative AI use; the firm issued a partial refund to the Australian government.

The prominent Wall Street law firm is Sullivan & Cromwell which apologized to a federal bankruptcy judge in April 2026 after submitting an emergency motion containing dozens of fabricated case citations and AI-generated errors.

Meta's AI chatbot falsely told users that conservative activist Robby Starbuck had participated in the January 6, 2021 attack on the U.S. Capitol. Starbuck sued Meta for defamation over the AI-generated hallucination; the two sides reached a settlement in August 2025, after which Starbuck advised Meta on AI bias.