In an embarrassing turn of events for the professional services industry, KPMG has been forced to retract a high-profile white paper titled “Redefining excellence in the age of agentic AI.” The report, which was intended to position the firm as a thought leader in the rapidly evolving landscape of artificial intelligence, instead became a cautionary tale of the dangers inherent in the technology itself.
The retraction followed a wave of pushback from major organizations cited within the document, who alleged that the report contained fabricated data and entirely false claims regarding their internal adoption of AI tools. The incident, first brought to public attention by the research group GPTZero, has ignited a broader conversation about the risks of "AI-generated journalism" within the corporate sector, suggesting that firms may be outsourcing their intellectual output to the very technologies they are attempting to audit.
The Anatomy of the Inaccuracies
The report, published in October 2025, was designed to showcase the transformative power of "agentic AI"—AI systems capable of performing complex tasks with minimal human intervention. However, the document’s credibility collapsed shortly after its release.
GPTZero, an organization specializing in AI detection and analysis, performed a deep-dive investigation into the document’s claims. Their findings were stark: the report was riddled with "hallucinations"—a phenomenon where AI models generate plausible-sounding but entirely fictional information. It appears that in its haste to capitalize on the AI trend, KPMG may have utilized AI tools to draft the report, effectively creating a feedback loop where an AI system hallucinated about the capabilities of other AI systems.
The scope of the errors was not limited to minor typos or formatting issues; the report attributed specific, high-level AI deployment strategies to organizations that had no such programs in place.
Chronology: From Publication to Retraction
- October 2025: KPMG officially publishes “Redefining excellence in the age of agentic AI,” promoting it as a flagship piece of research.
- Late October 2025: Independent researchers, led by GPTZero, begin vetting the report’s citations and case studies, noticing patterns consistent with LLM (Large Language Model) output.
- Early November 2025: High-profile institutions mentioned in the report, including UBS and the UK’s National Health Service (NHS), flag the content as inaccurate to media outlets, specifically the Financial Times.
- Mid-November 2025: KPMG issues a formal statement confirming the removal of the report from its digital channels and initiates an internal investigation.
Challenging the Claims: A List of Disputed Narratives
The impact of the report’s inaccuracies hit several major global entities, forcing them to issue public denials. The claims regarding their AI usage were described by spokespeople as "untrue," "misleading," or "fictional."
UBS
The Swiss multinational investment bank was cited as having implemented specific agentic AI workflows for risk management. UBS representatives confirmed that while they have an active AI strategy, the specific claims made in the KPMG report did not align with their internal technology roadmap or current operational reality.
The National Health Service (NHS)
The NHS, which has been under significant scrutiny regarding its digital transformation efforts, was portrayed as having integrated AI agents for patient diagnostics at a scale that simply does not exist. The NHS clarified that such claims were unauthorized and factually incorrect.
Swiss Federal Railways and Transport for London (TfL)
Both transport giants were credited with AI-driven operational efficiency projects that they claim never took place. These errors were particularly damaging, as they painted a picture of infrastructure automation that had not been vetted or approved by the respective governing boards of these organizations.
Official Responses and Internal Accountability
Following the backlash, KPMG moved quickly to distance itself from the document while acknowledging the gravity of the oversight. A spokesperson for the firm stated:
"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. We have removed the report from our websites while conducting our own investigation into how this occurred."
While the statement emphasizes "human oversight," it inadvertently raises the question: where was that oversight during the drafting and editorial phases of the publication? The incident suggests a failure in the firm’s internal quality control protocols, which are supposed to serve as the bedrock of professional services integrity.
The Broader Context: A Pattern of Failure
This is not an isolated incident. The KPMG scandal comes on the heels of another high-profile failure in the consulting sector. Just last month, EY (formerly Ernst & Young) was forced to withdraw its own report concerning loyalty rewards programs.
In that instance, the report appeared to contain "ghost footnotes"—references to non-existent academic papers or studies—which were traced back to hallucinations produced by generative AI. When taken together, these two events signal a systemic issue within the "Big Four" and similar professional services firms. There is an immense pressure to produce content at the speed of the AI news cycle, leading firms to prioritize speed and volume over the painstaking, manual verification process that once defined their reputation.
Implications: The Death of Expert Credibility?
The implications of these failures are profound. For firms like KPMG and EY, the primary product they sell is "authority." Clients pay millions for reports, white papers, and strategic advice because they trust the rigor of the research behind it.
1. Erosion of Brand Trust
When a firm publishes AI-generated hallucinations, they are not just making a factual error; they are eroding the fundamental contract of trust with their clients. If an audit report or a strategic analysis is found to contain AI-generated fabrications, how can a client trust the firm’s financial reporting or risk assessments?
2. The Liability of Automated Content
There is a burgeoning legal debate regarding the liability of firms that use AI. If a client makes a strategic business decision based on an AI-hallucinated report provided by their consultant, who is liable for the losses? This incident creates a legal gray area that will likely lead to stricter contractual clauses regarding the use of AI in professional deliverables.
3. The Need for "Human-in-the-Loop" Rigor
The industry is now facing a mandatory pivot back to traditional editorial standards. The allure of using AI to summarize data, draft sections, or generate charts must be tempered by a rigorous, "human-in-the-loop" verification process. GPTZero’s success in flagging these errors proves that the tools to detect AI fakery are evolving as fast as the tools to create it.
4. Regulatory Scrutiny
It is highly probable that regulators will begin looking into the "AI-generated content" policies of professional service firms. If these firms are performing "AI-assisted research," they may soon be required to provide transparency disclosures, detailing which parts of a document were generated by machines and which were authored by human experts.
Conclusion: Lessons for the AI Age
The KPMG incident serves as a stark reminder that in the age of agentic AI, the human role has not been diminished—it has become more critical than ever. The irony of a professional services firm failing to understand the limitations of the very tools they are advising others to adopt is not lost on the market.
As we move deeper into 2026, the firms that will survive and thrive are those that treat AI as a junior assistant rather than a senior partner. The "hallucination era" of corporate literature must come to an end if these firms wish to maintain their status as the guardians of corporate truth. For now, KPMG must grapple with the fallout, re-examine its internal workflows, and work to regain the trust of the clients it inadvertently misrepresented.
The promise of AI to "redefine excellence" remains, but as this episode demonstrates, true excellence still requires the one thing an algorithm cannot provide: accountability.
