KPMG retracts AI report after hallucinations surface in claims about organizations

Professional services giant KPMG has withdrawn a widely circulated report on artificial intelligence after multiple organizations named in the document disputed its accuracy, with independent researchers attributing the errors to AI hallucinations.

Report on agentic AI pulled after fact-checking reveals inaccuracies

The report, titled “Redefining excellence in the age of agentic AI,” was published in October 2025 and promoted as an analysis of how organizations are adopting advanced AI systems. However, in recent weeks, several entities cited in the report—including UBS, the UK’s National Health Service, Swiss Federal Railways, and Transport for London—told the Financial Times that the claims made about their AI usage were either untrue or misleading.

Research group GPTZero, which specializes in detecting AI-generated content and factual errors, identified numerous inaccuracies within the report. According to GPTZero, these errors appeared to stem from AI hallucinations, where a language model generates plausible-sounding but false information. The implication is that KPMG itself may have used AI tools to help write the report, without adequate human verification.

KPMG responds, launches internal investigation

A KPMG spokesperson confirmed the report has been removed from the firm’s websites while an internal investigation is conducted. “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,” the spokesperson said in a statement.

The incident is the second high-profile retraction by a major professional services firm in recent weeks. Last month, EY withdrew a report on loyalty rewards programs that appeared to include fake footnotes and AI-generated hallucinations, raising broader questions about the industry’s reliance on generative AI for research and analysis.

Why this matters for the professional services industry

KPMG’s retraction underscores a growing risk for consulting and advisory firms that increasingly use AI to produce client-facing reports. The incident highlights the difficulty of ensuring factual accuracy when generative AI tools are involved, particularly when the output is not rigorously fact-checked by human experts. For clients and regulators, it raises concerns about the reliability of AI-assisted research and the potential for reputational damage when errors are exposed.

The case also adds to the ongoing debate about transparency in AI usage. If firms like KPMG are using AI to generate reports about AI adoption, the irony is not lost on industry observers, and it may accelerate calls for clearer disclosure requirements and stronger verification protocols.

Conclusion

KPMG’s decision to pull its report on agentic AI, following the discovery of apparent AI hallucinations, represents a significant credibility challenge for the firm and the broader professional services sector. As the industry grapples with how to responsibly integrate generative AI into its workflows, this incident serves as a cautionary tale about the necessity of human oversight and source verification. The outcome of KPMG’s internal investigation will be closely watched by clients, competitors, and regulators alike.

FAQs

Q1: What was the KPMG report about?
The report, titled “Redefining excellence in the age of agentic AI,” was published in October 2025 and discussed how organizations are using advanced AI systems. It was withdrawn after inaccuracies were found.

Q2: What is an AI hallucination?
An AI hallucination occurs when a language model generates information that sounds plausible but is factually incorrect or fabricated. In this case, GPTZero identified hallucinations as the source of the report’s false claims.

Q3: Which organizations disputed the report’s claims?
UBS, the UK’s National Health Service, Swiss Federal Railways, and Transport for London all told the Financial Times that the report’s statements about their AI usage were untrue or misleading.