KPMG’s much-discussed October 2025 report on agentic AI has come under fire after a review found that only a small fraction of its citations matched actual sources, highlighting widespread inaccuracies and fabricated references within the document.
- Only 5 of 45 citations in KPMG’s AI report verified by GPTZero
- Numerous case studies and statistics found to be false or unsupported
- KPMG removes report pending investigation into its publication process
What happened
KPMG released a report in October 2025 entitled "Total Experience: Redefining Excellence in the Age of Agentic AI," claiming insights into cutting-edge AI applications. However, the investigative group GPTZero conducted a detailed review and found substantial issues with the report’s references. Out of 45 citations, only 5 correctly matched their sources; the rest were either distorted, vague, or invented, a phenomenon dubbed "vibe citing."
GPTZero also exposed falsehoods in case studies cited to demonstrate agentic AI deployments, including exaggerated claims about organizations like UBS, Swiss Federal Railways, Transport for London, and Emirates Airlines. Some data points even conflicted with KPMG's own research published in the same month. In response, KPMG has removed the report from certain digital platforms and initiated an internal review of its development and release.
Why it matters
This incident underscores the risk that even major consulting firms face when relying on generative AI for content creation without sufficient human oversight and verification. The presence of fabricated or inaccurate citations not only undermines the credibility of the report but also raises concerns about the quality and accountability of AI-assisted publishing in professional services.
KPMG’s case highlights the ongoing challenge of verifying AI-generated outputs, especially when they involve complex references and data claims. As organizations increasingly integrate AI in producing research and thought leadership, ensuring rigorous validation becomes critical to maintain trust and reliability.
What to watch next
Observers will follow KPMG’s investigation progress and whether the firm will implement stronger safeguards and new protocols for AI-generated content. This case may prompt other consultancies and knowledge-based industries to reassess their quality control measures amid growing AI integration.
Additionally, stakeholders in AI ethics and policy may use this example to argue for clearer standards or guidelines on the responsible deployment of AI in research publications, particularly regarding source verification and factual accuracy.