KPMG pulls AI report after clients say claims about their AI use were false
KPMG pulls AI report that cited client AI deployments after multiple organizations disputed its claims, prompting the firm to remove the October 2025 paper and launch an inquiry.
KPMG pulls AI report after clients dispute claims
KPMG has removed a report titled “Redefining excellence in the age of agentic AI” from its websites after several organizations told the Financial Times that claims in the paper about their use of artificial intelligence were incorrect or misleading. The report, published in October 2025, was flagged for factual errors by researchers and clients, and an external group attributed many issues to AI-generated inaccuracies. The withdrawal follows mounting scrutiny of how consultancies and research teams are using generative AI tools in public-facing analysis.
Allegations of AI-generated inaccuracies raised by researchers
GPTZero identified multiple inaccuracies in the KPMG report and told the Financial Times that those errors appeared to stem from AI hallucinations. Researchers said the paper contained assertions about specific organizations’ AI deployments that could not be corroborated by the named entities. The identification of hallucinated details prompted broader questions about editorial controls when firms use generative models to draft analytical reports.
GPTZero’s findings are notable because the group specializes in detecting synthetic or machine-generated content. Its assessment suggested that portions of the report were likely drafted or assisted by AI without sufficient human validation. That assessment has pushed KPMG to reassess its vetting procedures for AI-assisted research.
Clients named in the report disputed key claims
Multiple organizations named in the report — including UBS, the UK’s National Health Service, Swiss Federal Railways and Transport for London — told the Financial Times that the report’s descriptions of their AI usage were untrue or misleading. Each organization said the paper overstated or misstated how it deploys AI, and some noted that the report implied existing programs or partnerships that do not exist.
These denials were central to the decision to pull the report, according to sources familiar with the matter. The reactions from high-profile public and private institutions highlighted the reputational risk for consultancies that publish analyses asserting client practices without clear, independently verifiable evidence.
KPMG begins an internal investigation and reaffirms oversight rules
A KPMG spokesperson said the firm removed the report while it conducts an internal investigation into the document’s production and the accuracy of its claims. The spokesperson emphasized that KPMG expects staff to follow established guidelines on the responsible use of AI, including human oversight to validate content and verify independent sources.
KPMG’s statement did not specify whether the report was drafted with the assistance of specific generative models or name the teams involved. The firm said it is reviewing editorial controls and compliance with its internal AI policies, suggesting potential changes to how future research is produced and reviewed.
Industry context: recent withdrawals and heightened scrutiny
The KPMG episode follows a recent instance in which another Big Four firm, EY, withdrew a separate report after apparent AI-related errors were discovered. Media reporting noted that EY’s withdrawn document contained fake footnotes and instances of fabricated sourcing, prompting that firm to remove the material and conduct its own review.
Taken together, the two incidents have intensified scrutiny of how major professional services firms incorporate generative AI into research and client outputs. Regulators, clients and independent researchers are increasingly demanding transparency about whether and how AI tools are used, and what safeguards are in place to prevent fabrications or misattribution.
Calls for clearer verification standards and public corrections
Newsroom and academic observers have called for stronger verification standards when firms publish research referencing organizations’ practices. Analysts recommend that consultancies adopt mandatory source trails, human-in-the-loop sign-offs, and explicit disclosures when content has been assisted or produced by AI. These measures, proponents say, would reduce the risk of publishing inaccurate claims and protect both the named organizations and the firms producing the work.
Public-interest groups and clients have also urged firms to issue corrections and make investigative findings public where third parties are potentially misrepresented. In the current case, the named organizations have already publicly disputed the report’s assertions, increasing pressure on KPMG to disclose the findings of its internal inquiry.
The development adds to an ongoing debate about the limits of generative AI in professional research and the responsibilities of large consultancies to maintain rigorous fact-checking. As firms continue to experiment with agentic AI and automated drafting tools, the incident underscores the potential reputational and legal consequences of unchecked AI assistance.
KPMG’s removal of the October 2025 report and its pledge to investigate provide a test case for industry practices, and the outcome of the firm’s review is likely to influence how consultancies disclose AI usage in future publications.
