KPMG has removed its October 2025 report on AI usage after several organizations disputed the accuracy of its findings, citing instances of AI hallucinations in the document.
- KPMG retracts AI usage report over factual inaccuracies
- Multiple organizations disputed claims made in the report
- AI hallucinations identified as the root cause of errors
What happened
KPMG has pulled its October 2025 report titled 'Redefining excellence in the age of agentic AI' after numerous organizations flagged the report’s claims about their AI usage as false or misleading. The report included statements regarding AI practices at UBS, the UK’s National Health Service, Swiss Federal Railways, and Transport for London, all of which disputed the accuracy of the report’s assertions.
Independent research by GPTZero revealed that the incorrect claims were due to AI hallucinations—fabricated or erroneous content generated by AI systems. In response, KPMG temporarily removed the report from their platforms and initiated an internal investigation to assess the origins and extent of these inaccuracies.
Why it matters
This incident highlights the risks of over-reliance on AI-generated content in professional and authoritative reports. KPMG’s case illustrates how AI hallucinations can undermine credibility and lead to the dissemination of false information, even in high-profile, expert-driven publications.
It also underscores the importance of human oversight and source verification when integrating AI into research and reporting processes. As firms increasingly adopt AI tools, strict adherence to guidelines on responsible AI use becomes essential to maintain trust and accuracy.
What to watch next
KPMG’s ongoing investigation will likely shed light on how the AI hallucinations occurred and whether procedural changes are needed to prevent future errors. The outcome may influence broader professional services industry protocols for AI integration and content validation.
Additionally, the incident may prompt other firms to scrutinize AI-generated content more carefully and enforce stronger editorial controls. Parallel examples, such as EY’s recent report withdrawal due to AI hallucinations, indicate an emerging pattern of challenges tied to AI-driven research.