KPMG pulls its agentic AI report after GPTZero finds 40 of 45 citations were wrong

KPMG has pulled its report "Total Experience: Redefining Excellence in the Age of Agentic AI" after multiple organisations named in it said its claims about their AI deployments were false.

An investigation by GPTZero found that 40 of the report's 45 citations were wrong. Only five pointed accurately to real sources. Another 28 had paraphrased or fabricated components; the remaining 12 were too vague to verify at all. UBS, the UK's National Health Service, Swiss Federal Railways, and Transport for London all told the Financial Times that the report's descriptions of their agentic AI use were either untrue or misleading. A KPMG spokesperson said the firm has removed the report from its websites while conducting its own investigation.

The mechanism GPTZero identified was AI hallucination: an AI tool appears to have conflated sources, invented case study details, and injected references to agentic AI that were then carried into the final report without human verification. The irony is hard to miss. A major professional services firm produced a report on the transformative benefits of agentic AI, and the most agentic part of the process was quietly fabricating 89% of the citations.