KPMG Pulls AI Usage Report After Hallucinations Attributed to Artificial Intelligence Tools

Professional services firm KPMG has removed a report on artificial intelligence from its websites after multiple organizations named in the document said its claims about their AI usage were inaccurate or misleading, highlighting the risks firms face when deploying AI to produce research about AI.

The report, titled "Redefining excellence in the age of agentic AI," was originally published in October 2025. Research group GPTZero identified a series of inaccuracies in the document and told the Financial Times that the errors stemmed from AI hallucinations — instances where AI systems generate plausible-sounding but fabricated information.

UBS, the United Kingdom's National Health Service, Swiss Federal Railways, and Transport for London all informed the Financial Times that statements in the report concerning their AI adoption were either untrue or misleading.

A KPMG spokesperson confirmed the removal and said the firm is conducting its own investigation. "We expect all our people to follow our guidelines on the responsible use of AI, including human oversight to validate content and verify independent sources," the spokesperson said.

The episode draws attention to a growing pattern among professional services firms. Last month, rival firm EY withdrew a separate report on loyalty rewards programs that appeared to contain fabricated footnotes and AI-generated inaccuracies.

For KPMG, the retraction carries particular weight given the report's subject matter — the firm appears to have used AI to help author a document whose central purpose was to advise organizations on how to deploy AI responsibly and effectively.

The incidents at both KPMG and EY underscore a tension facing consulting and advisory firms as they race to position themselves as AI authorities while simultaneously integrating AI tools into their own production workflows. The reputational stakes are especially high when clients and the organizations cited in such reports can directly contradict the published claims.

GPTZero, which developed tools for detecting AI-generated content, has become an increasingly prominent actor in flagging such errors in professional and institutional documents, reinforcing demand for independent verification mechanisms even within firms that market their own AI governance frameworks.

How KPMG's investigation concludes — and whether the report is revised, replaced, or permanently withdrawn — will be watched closely by clients and competitors navigating similar risks in their own AI-assisted research pipelines.