KPMG pulled a report called “Redefining excellence in the age of agentic AI” after clients flagged false claims about their AI usage. GPTZero traced many errors to AI hallucinations, revealing the firm had likely used AI to draft a piece about AI itself. The move follows similar fallout for EY last month.

Clients such as UBS, the UK National Health Service, Swiss Federal Railways and Transport for London said the report misrepresented their AI adoption. KPMG said it removed the document while conducting an internal review and reiterated its policy that all staff must validate content with human oversight. The episode underscores the risks of auto‑generated executive briefs.

The incident follows EY’s earlier withdrawal of a loyalty‑rewards report that contained fabricated footnotes and hallucinated data. Both cases show that even top consulting houses can fall prey to the very tools they promote. Firms must tighten verification processes or face reputational damage that can cost millions in client trust.

KPMG’s withdrawal signals a broader industry reckoning, as regulators and clients demand higher standards for AI‑generated content. The firm’s spokesperson warned that all employees must adhere to strict AI‑use guidelines, a reminder that human oversight remains essential when deploying generative models in high‑stakes environments.