4 min readNew DelhiJun 15, 2026 09:06 AM IST
First published on: Jun 14, 2026 at 12:06 PM IST
KPMG has withdrawn a study on agentic AI that reportedly included AI-generated hallucinations and fake citations, in the latest example of a professional services firm being caught by the shortcomings of the technology.
The report titled ‘Total Experience: Redefining Excellence in the Age of Agentic AI’ was published in October 2025. However, research group GPTZero recently identified a number of inaccuracies in the report, many of which stemmed from AI-generated hallucinations. This means that the company seems to have used AI tools to draft the report about agentic AI.
KPMG pulled the report after several organisations mentioned in the document said that its claims about their AI usage were untrue. UBS, the UK’s National Health Service, Swiss Federal Railways, and Transport for London said that the KPMG report’s claims about their AI usage were either untrue or misleading, according to Financial Times.
The incident underscores how consulting firms have become particularly vulnerable to AI-generated errors, even as they caution clients about safeguarding against AI hallucinations and other reliability issues. It also comes just one month after EY withdrew a report on loyalty rewards programs that appeared to include fake footnotes and AI hallucinations.
Last year, Deloitte was forced to refund the Australian government after AI-generated content made its way into a taxpayer-funded report.
“KPMG International takes the accuracy and integrity of its published content seriously. The report has been removed and we are reviewing the circumstances surrounding its publication. 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,” a company spokesperson was quoted as saying by The Register.
What did researchers find?
Upon conducting a forensic review of KPMG’s report, GPTZero researchers said that they found only five out of 45 citations in the report correctly pointed to the cited source. The remaining citations were either misleading, partially fabricated, or too vague to verify. The trend of generative AI tools making up citations by stitching together pieces of real sources or inventing entirely new ones has been dubbed as ‘vibe citing’ by GPTZero.
Additionally, nearly half of the report’s claims were false, unsupported, or attributed to the wrong source. Several studies mentioned in the report that looked to highlight cutting-edge deployments of AI agents were also found to be creative, as per the researchers.
Case studies about purported agentic AI deployments at organisations such as UBS, Swiss Federal Railways, and Transport for London either failed to substantiate the report’s claims or contained alterations and paraphrasing that undermined their reliability, they said.
“These factual errors are not confined to the report’s footnoted passages. On page 42, the authors claim that Emirates airline has adopted a mobile chatbot named Sara (false) that can converse directly with passengers (partially true) and change their flights (false). In fact, Sara is a robot assistant introduced by Emirates in 2023 (not a chatbot) that lacks the ability to alter flight bookings,” GPTZero researchers said.
The research firm further noted that certain figures appeared to contradict the firm’s previous reports. For instance, a statistic that said over 55 percent of CEOs ranked AI as their top investment priority was in contradiction with the 71 per cent figure mentioned in KPMG’s 2025 CEO Outlook that came out during the same month.
GPTZero researchers said that they found only five out of 45 citations in the report correctly pointed to the cited source. (File photo) 
Share
