KPMG Withdraws AI Report With Hallucinated Citations and False Case Studies
KPMG withdrew a report on agentic AI after research by GPTZero revealed that 40 of 45 citations appeared to be AI hallucinations and multiple case studies about companies including UBS, NHS Greater Manchester, and Transport for London were factually inaccurate or misleading. The report was widely referenced in publications and AI chatbots before removal.
What happened
KPMG published a report titled 'Redefining excellence in the age of agentic AI' in October that contained multiple false case studies about companies using AI agents. According to GPTZero, the report included 40 of 45 citations that appeared to be at least partially AI hallucinations, and GPTZero flagged several passages as hallucinations which were verified by the Financial Times.
The report made false claims about UBS integrating AI agents in wealth management and compliance monitoring with Microsoft. It stated Swiss Federal Railways deployed AI agents for journey planning and optimization, which the company said was 'not accurate'. It claimed Transport for London uses AI agents for congestion prediction and management, which TfL said was 'misleading'. It asserted NHS Greater Manchester uses AI agents to predict readmissions and triage patients, which a spokesperson said 'doesn't really align' with the press release cited. The report also falsely claimed Japanese East Japan Railway Company uses agentic AI for customer service based on a 2019 press release, years before agentic AI became mainstream.
Reported impact
The report was referenced by multiple customer service publications and a major newspaper in the Czech Republic. Researchers noted the paper is being referenced as a resource on customer experience by OpenAI's ChatGPT and Google Gemini, potentially spreading misinformation.
GPTZero alleged that 'no human at KPMG double-checked the citations, the claims, or the sources' before publication. GPTZero CEO Edward Tian said error-filled reports from the Big Four 'poison the well of information'.
Index analysis
GPTZero suspects an LLM research tool was used to find case studies of companies using agentic AI worldwide, which 'seems to have conflated sources, exaggerated claims, and injected references to agentic AI' without verification.
Many citations in the report linked to existing studies but had incorrect titles, authors, or dates, while researchers were unable to match many other citations to existing studies due to insufficient information.
Governance controls
KPMG removed the report from its websites and stated it is reviewing how it was produced. A KPMG spokesman said the firm 'takes the accuracy and integrity of its published content seriously' and that it 'expects all our people to follow our guidelines on the responsible use of AI, including human oversight to validate content and verify independent sources.'
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.