The news: KPMG pulled its October 2025 “Redefining excellence in the age of agentic AI” report after multiple organizations said its claims about their AI use were false, per the Financial Times.

Several subjects of KPMG’s report, including Swiss Federal Railways, Transport for London, UBS, and the UK National Health Service, disputed some findings as false or misleading, while research firm GPTZero told FT that the report’s errors were due to AI hallucinations.

“Of the 45 citations in the report, only five accurately point to real sources. Another 28 citations provide paraphrased titles and/or fake components for a real source,” GPTZero noted in an article. 

Errors in consulting firm data can have wide reach, including within large language models (LLMs) that may have been trained on that data and for marketers who rely on those insights for advertising and brand decisions.

Zooming out: KPMG’s report wasn’t an isolated case. GPTZero’s head of machine learning, Alex Adams, calls the issue “vibe citing,” or when AI models create bogus references through LLM hallucinations. 

Last month, EY withdrew a loyalty report with fake footnotes, per the Financial Times. GPTZero revealed that “more than half a dozen of the footnotes in EY’s report directed to web pages that did not exist or did not contain the information cited.”

In October, Deloitte Australia partially refunded the Australian government for a report with alleged AI-generated errors, per the Associated Press. A revised version of the report included a disclosure that stated Azure OpenAI was used in it.

Why it’s worth watching: These incidents trigger a trust crisis around AI’s use for research and the probability of undetected hallucinations being published.

• KPMG’s report was in circulation for roughly eight months before getting pulled.
• Three of the Big Four consulting firms using AI-generated information without checking it in a matter of months underscores the growing risk in automating research without stricter oversight.

Implications for brands: When consulting firms publish reports with inaccurate or hallucinated data, the reputational fallout doesn’t stop with them—once brands circulate and repeat that data, it triggers a cascade of bad decisions, wasted ad spend, and broken trust.

• Brands should require AI disclosures that include when consultants use genAI for research and to what extent. 
• Build a rapid-response plan for when false statements about your brand appear in paid research or news or are surfaced by LLMs. This could involve issuing a public statement, contacting the publisher for retraction, and cross-verification of all cited claims.

Unchecked hallucinations create a burden no brand asked for. Due diligence now means treating vendor reports as drafts until every claim is verified—especially because AI models train on published material, amplifying errors and hallucinations at scale.

This content is part of EMARKETER’s subscription Briefings, where we pair daily updates with data and analysis from forecasts and research reports. Our Briefings prepare you to start your day informed, to provide critical insights in an important meeting, and to understand the context of what’s happening in your industry. Non-clients can click here to get a demo of our full platform and coverage.