KPMG has taken down a report about artificial intelligence after several organizations named in it said the document described their AI use incorrectly. The report, titled Redefining excellence in the age of agentic AI, was published in October 2025 and has since drawn scrutiny over what appear to be AI-generated errors.
The issue came to light after research group GPTZero identified a number of inaccurate claims in the report. GPTZero told the Financial Times that the mistakes were likely caused by hallucinations, a term used to describe false or invented output from AI systems. The findings raised the possibility that AI tools may have been used in the preparation of a report that was itself about AI.
Several major organizations named in the document said KPMG’s descriptions of their AI adoption were wrong or misleading. UBS, the U.K.’s National Health Service, Swiss Federal Railways, and Transport for London all told the Financial Times that the report overstated, misstated, or otherwise misrepresented their AI use.
A KPMG spokesperson said the firm had removed the report from its websites while it carries out its own review. The spokesperson also said KPMG expects employees to follow internal rules on responsible AI use, including human oversight to verify content and confirm independent sources.
The incident comes at a time when professional services firms are under increasing pressure to ensure that AI-assisted work is accurate and properly checked. The report’s removal is a reminder that even companies advising clients on emerging technology are still vulnerable to the shortcomings of generative AI.
The episode also follows a similar case involving another major accounting firm. Last month, EY withdrew a report on loyalty rewards programs after concerns that it contained fake footnotes and apparent AI hallucinations. Together, the two incidents suggest that the problem is not limited to experimental projects or consumer-facing tools, but can also affect formal research and consulting materials.
KPMG has not publicly detailed which parts of the report were incorrect or whether the mistakes were made by an internal team, an outside contractor, or an AI tool used during drafting. The firm’s review is ongoing.
The broader concern for businesses is not simply that AI can produce wrong answers, but that those errors can be embedded in polished, authoritative-looking documents. In this case, the report was presented as a study of how organizations are using AI, yet several of those organizations said the findings did not reflect reality.
For companies deploying AI in research, consulting, and marketing, the case underscores the need for stronger verification steps before publication. As KPMG’s own response indicates, human review remains central to catching errors that automated systems may introduce.
The report’s withdrawal may also fuel wider skepticism about AI-generated business intelligence, especially when the material is used to make claims about large institutions and public-sector organizations. With firms increasingly racing to adopt AI tools, the challenge is not just using them, but proving the information they produce is trustworthy.