In a deeply ironic twist that underscores the very problem it sought to address, professional services giant KPMG has been forced to retract a report on artificial intelligence after it was found to contain multiple fabricated case studies — a classic example of AI hallucinations.
How a report on AI became a cautionary tale about AI
The report, titled "Redefining excellence in the age of agentic AI," was published in October 2025 and aimed to showcase how leading organizations were adopting autonomous AI agents. But instead of demonstrating KPMG's expertise, it became an accidental demo of why AI cannot yet be trusted without human verification.
Research firm GPTZero, which specializes in detecting AI-generated content and inaccuracies, identified numerous problems. According to their analysis, only 5 of the report's 45 citations actually matched their claimed sources. The rest appeared to be hallucinations — confident but false outputs generated by AI systems.
Why this matters for businesses relying on AI consulting
For companies paying top dollar for Big Four consulting advice, this incident raises uncomfortable questions. If KPMG cannot ensure accuracy in its own AI report — a document meant to demonstrate thought leadership — how can clients trust its recommendations on implementing AI systems?
The financial stakes are enormous. Businesses globally are pouring billions into AI adoption, often relying on consultants like KPMG, Deloitte, EY, and PwC for guidance. A report that fabricates case studies undermines the entire consulting value proposition.
What the report actually claimed — and what was wrong
The report included case studies claiming that UBS, along with various health and transit systems, were using KPMG's AI tools in specific ways. Multiple organizations contacted KPMG to say these claims were simply untrue. The report did not just exaggerate — it invented specific use cases that never existed.
According to TechCrunch's reporting, which first broke the story, the inaccuracies were not minor errors but fundamental fabrications about how organizations were deploying AI. This is the hallmark of AI hallucination: the system generates plausible-sounding but entirely fictional information.
Who is affected by this retraction
The immediate victims are the organizations falsely named in the report, who now face questions from stakeholders about AI practices they never implemented. But the ripple effects extend to KPMG's entire client base, particularly those in financial services and healthcare who rely on the firm's AI expertise.
For the broader consulting industry, this is a reputational blow. If a Big Four firm cannot verify basic facts in a flagship report, it raises systemic questions about quality control across the sector.
KPMG's response — or lack thereof
As of the latest reports, KPMG has not issued a detailed public statement explaining how the hallucinations occurred or what internal processes failed. The report has simply been removed from the company's website. This silence is itself telling — and damaging.
Industry observers note that the lack of transparency could compound the reputational harm. In an era where AI accountability is a hot-button issue, a Big Four firm's refusal to explain its own AI failure sends a troubling signal.
The deeper irony: AI's unreliability about AI
Perhaps the most striking aspect of this story is that KPMG's report was about agentic AI — autonomous systems that can act without human intervention. The report's own hallucinations demonstrate precisely why such systems remain risky. AI cannot yet be trusted to accurately report on itself, let alone make autonomous decisions in high-stakes environments.
This is not an isolated incident. Multiple studies have shown that large language models frequently hallucinate when asked to summarize or report on technical topics, especially about other AI systems. The KPMG case is a high-profile example of a systemic problem.
Confirmed facts vs what remains unclear
Confirmed: KPMG pulled the report after organizations denied the claims. GPTZero found only 5 of 45 citations matched sources. The report was published in October 2025 and withdrawn in June 2026.
Unclear: Whether KPMG used AI to generate the report itself, or whether human researchers simply failed to verify AI-generated content. The exact internal process that led to the hallucinations has not been disclosed.
Speculation: Some analysts suggest the report may have been partially AI-generated, which would explain the scale of hallucinations. This has not been confirmed by KPMG.
KPMG's competitive position in AI consulting
KPMG has invested heavily in AI capabilities, positioning itself as a leader in helping clients adopt generative AI and agentic systems. The firm has partnerships with major AI providers and has launched its own AI tools for audit and advisory work.
However, this incident threatens to undermine that positioning. In the competitive Big Four landscape, trust is the primary differentiator. A firm that cannot produce an accurate report on its own area of expertise faces an uphill battle convincing clients of its competence.
Risks and balanced view
Critics argue that this incident reveals a deeper problem: consulting firms are rushing to monetize AI without adequate safeguards. Supporters of KPMG might counter that the firm acted responsibly by retracting the report once errors were identified, and that no consulting firm is immune to such mistakes.
The balanced view is that while the retraction is embarrassing, it is also an opportunity for the industry to implement better verification processes. The real failure would be if KPMG does not learn from this and improve its quality controls.
Wider trend: AI hallucinations in professional services
KPMG is not alone. Law firms have been caught filing court documents containing AI-hallucinated case citations. Medical researchers have published papers with fabricated references. The pattern is clear: as AI tools become ubiquitous, the risk of undetected hallucinations grows.
Professional services firms face a particular challenge because their value proposition rests on accuracy and trust. An AI hallucination in a consulting report is not just an error — it is a breach of the fundamental promise that clients are paying for expert judgment.
What businesses should do now
For organizations using or considering AI consulting services, this incident offers several lessons. First, always verify case studies and claims independently. Second, ask consulting firms about their AI verification processes. Third, consider requiring contractual guarantees about the accuracy of AI-generated content in deliverables.
For KPMG clients specifically, this may be a moment to request a review of any AI-related work the firm has delivered, and to seek clarity on how the firm will prevent similar incidents in the future.
What happens next
KPMG faces a choice: issue a detailed mea culpa with process improvements, or hope the story fades. Given the prominence of the incident — covered by TechCrunch, The Register, and the Financial Times — the former seems wiser.
Regulators may also take notice. If consulting firms cannot reliably produce accurate AI reports, it could trigger calls for greater oversight of AI advisory services, particularly in regulated industries like banking and healthcare.
Our Take
The KPMG report retraction is more than an embarrassing gaffe — it is a textbook case of the AI trust problem. If the very firms advising businesses on AI cannot avoid its pitfalls, the technology's adoption faces a credibility crisis. The irony is almost too perfect: a report on AI reliability became unreliable because of AI. The lesson for every business is simple: trust, but verify — especially when AI is involved.
Frequently Asked Questions
Why did KPMG pull its AI report?
KPMG retracted the report "Redefining excellence in the age of agentic AI" after multiple organizations said the report's claims about their AI usage were untrue. GPTZero found that only 5 of 45 citations matched their sources, indicating widespread hallucinations.
What is an AI hallucination?
An AI hallucination occurs when an artificial intelligence system generates information that is confident-sounding but factually incorrect or entirely fabricated. In KPMG's case, the report included case studies about companies using AI in ways that never happened.
Which organizations were falsely named in the KPMG report?
According to reports, the case studies included false claims about UBS, health systems, and transit systems using KPMG's AI tools. These organizations contacted KPMG to deny the claims.
Could this affect KPMG's consulting business?
Yes. Trust is the foundation of consulting relationships. This incident raises questions about KPMG's quality control processes and could lead clients to scrutinize the firm's AI advisory work more closely, potentially affecting future engagements.
Was the KPMG report itself written by AI?
KPMG has not confirmed whether AI was used to generate the report. However, the pattern of hallucinations — confident but false citations — is characteristic of AI-generated content that was not properly verified by human reviewers.
