AI failures · Fabricated content
KPMG Withdraws Agentic AI Report After Audit Finds Only 5 of 45 Citations Accurate
Verified Checked against its sources on 5 sources quoted
- Where it happened
- International or several countries
- Organization
- KPMG
- Date
- Report published October 2025; GPTZero investigation published June 12, 2026; removal first reported June 12, 2026
- Function
- Consulting / published research
- Where it happened
- Thought leadership publication
- System
- Not disclosed
- Technology
- Generative AI
- Type of failure
- Fabricated content
- Caught by
- A researcher
Summary
KPMG's October 2025 report 'Total Experience: Redefining Excellence in the Age of Agentic AI' was examined by GPTZero, which found that only five of its 45 citations accurately pointed to real sources; 28 carried paraphrased titles or invented components for a real source and 12 were too vague to trace. GPTZero also counted 40 of the 45 citation titles as fake, most of them paraphrases loosely matching real sources. About half of the claims the citations supported appeared fake or misattributed, including a 2019 JR East press release cited as evidence of present-day AI-agent use. GPTZero said the report's claims had already been repeated in trade publications and a Czech newspaper and were being cited by AI chatbots. UBS, NHS Greater Manchester, Swiss Federal Railways and Transport for London told the Financial Times the report's claims about their AI use were untrue or misleading. City AM reported on June 12, 2026, that KPMG had removed the report, and KPMG said it was reviewing how the report came to be published.
Consequence
- KPMG removed the report from its websites while it investigated.
Who caught it A researcherFound outside the organization
Analysis
What could have been done
Checking AI output against its sources
Every factual element of AI output (citations, figures, quotations) is checked against its source before reliance.
Only five of the report's 45 citations accurately pointed to real sources; 28 carried paraphrased titles or invented components and 12 were too vague to trace.
Evidence behind claims about AI
Public statements about AI capability are backed by evidence before they are made.
The report made public claims about the AI use of named organizations, and UBS, NHS Greater Manchester, Swiss Federal Railways and Transport for London told the Financial Times those claims were untrue or misleading.
Inputs and source data are complete, current and fit for the use before the AI acts on them.
A 2019 JR East press release was cited as evidence of present-day use of AI agents, so the source was not current for the claim it supported.
What would likely have prevented it
- Before publishing a statement that a named organization uses a technology, confirm it from a current primary source or with the organization itself.
- Check that each source's date and content fit the specific claim it is cited for, not only that the source exists.
What would likely have caught it earlier
- Before release, have a reviewer independent of the authors trace every claim about a named organization to its source, and hold publication if any cannot be traced.
Key dates
- June 12, 2026Event
- October 16, 2026Checked against sources
- October 18, 2026Checked against sources
- October 20, 2026Checked against sources
- October 21, 2026Checked against sources
Developments are added when the record is reviewed. Next review: October 20, 2027.
Evidence
Each passage is quoted from its source. The quoted passages are re-checked against the live pages every month.
-
GPTZero's investigation revealed only 5 of 45 citations were accurate, leading to misinformation and reputational harm for KPMG and affected organizations.
Supports: Incident logged; 5 of 45 citations accurate (the monitor marks this text as AI generated)
-
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. The final 12 are too vague or flawed to accurately determine if a source exists.
Supports: Citation breakdown
the report seems to cite a 2019 press release from the East Japan Railway Company (JR East) as evidence that the railway is leveraging AI agents to make travel recommendations and predict service disruptions.
Supports: 2019 press release cited as evidence of current agentic AI use
This investigation analyzes a KPMG report from October 2025 on customer experience and agentic AI.
Supports: Report publication date and subject
The task of citation verification is made even more challenging by citation titles: 40 of 45 are fake.
Supports: 40 of 45 counts citation titles, not citations
In these stories, claims, including the 55% statistic mentioned above, have been repeated and recycled
Supports: GPTZero: the claims were repeated in trade publications and a Czech newspaper
KPMG’s report is also being cited by LLMs, which can strip claims of context
Supports: GPTZero: AI chatbots cite the report
-
UBS, the U.K.’s National Health Service, Swiss Federal Railways, and Transport for London all told the FT that the report’s claims about their AI usage were either untrue or misleading. A KPMG spokesperson said the firm removed the report from its websites while conducting its own investigation.
Supports: Named organizations disputed claims; withdrawal
-
Friday 12 June 2026 6:17 am | Updated: Friday 12 June 2026 4:02 pm KPMG report on AI found riddled with AI hallucinations
Supports: removal reported on June 12, 2026 (page date line)
The report has been removed and we are reviewing the circumstances surrounding its publication.
Supports: KPMG spokesperson: report removed, circumstances under review
-
made several false claims about the use of AI at UBS, NHS Greater Manchester, Swiss Federal Railways and Transport for London (TfL)
Supports: the NHS body named is NHS Greater Manchester
Suggested citation. Eminence Scan, “KPMG Withdraws Agentic AI Report After Audit Finds Only 5 of 45 Citations Accurate”, Eminence Growth Solutions, last verified October 20, 2026, https://eminencegrowthsolutions.com/tools/eminence-scan/record/fail-097/.
Using our research: you may read and quote Eminence Scan with credit to “Eminence Scan, Eminence Growth Solutions” and a link; all other rights reserved. Terms: Using our research.
Spotted an error, or a development we have missed? Report a correction. How we check records: research methodology.