KPMG AI Report Hallucinations Exposed: FT Verification Shows Misleading Claims by Major Clients

KPMG, one of the Big Four consulting firms, published the report “Redefining excellence in the age of agentic AI” in October of the previous year. The document claimed that major organizations were successfully deploying AI agents in real operations. However, independent fact-checking by the Financial Times revealed that several key claims were factually incorrect or misleading. KPMG later removed the report from its websites, but not before it had already been cited by other media outlets. This case highlights the emerging risk of “second-hand hallucinations,” where AI-generated falsehoods embedded in authoritative reports spread further through citations.

📑Table of Contents

KPMG Report Claims on AI Agent Adoption
Financial Times Independent Verification Results
The Risk of Second-Hand Hallucinations and Information Pollution
Comparison of Report Claims and Verification Outcomes
Lessons for AI Adoption in Consulting Firms
Frequently Asked Questions
Conclusion

KPMG Report Claims on AI Agent Adoption

The report presented four concrete examples of AI agent usage by prominent clients:

UBS: AI agents integrated across investment advisory, risk management, and compliance monitoring.
Swiss Federal Railways (SBB): AI agents assisting users with journey planning, booking, and optimization based on preferences, real-time conditions, and carbon impact.
Transport for London: AI agents for congestion prediction, personalized commuter updates, and multimodal transport coordination.
NHS Greater Manchester: AI agents supporting patient triage, hospital readmission prediction, and medical workflow optimization.

These claims were positioned to illustrate the value of consulting services in the age of agentic AI and were subsequently referenced by industry publications and a major Czech newspaper. Sources: Futurism (June 2026) and Financial Times reporting.

Financial Times Independent Verification Results

Following a tip from GPTZero CEO Edward Tian, the Financial Times contacted the organizations mentioned. The responses were as follows:

UBS described the claims as “factually incorrect.” Swiss Federal Railways stated the description was “not accurate.” Transport for London called the portrayal “misleading.” The NHS Greater Manchester case was traced to a misinterpretation of a press release about a lung cancer detection AI tool; no agentic AI usage matching the report’s description was found.

KPMG removed the report after being notified, yet the document had already influenced downstream coverage. Source: Futurism article “Consulting Firms’ AI Report Was Loaded With Hallucinations.”

The Risk of Second-Hand Hallucinations and Information Pollution

GPTZero’s Edward Tian characterized the incident as “second-hand hallucinations,” warning that fabrications originating from a firm of KPMG’s stature can “poison the well of information.” When AI-generated content enters authoritative reports without sufficient human oversight, subsequent citations by media and other organizations amplify the misinformation. The episode coincides with reports of other large consultancies, such as McKinsey, deploying thousands of AI agents internally, raising questions about whether verification processes are keeping pace with adoption speed.

Comparison of Report Claims and Verification Outcomes

The table below contrasts the original claims with the Financial Times findings.

Organization	Report Claim	FT Verification Result
UBS	AI agents integrated in advisory, risk, compliance	Factually incorrect
Swiss Federal Railways (SBB)	Journey planning based on preferences, real-time, carbon	Not accurate
Transport for London	Congestion prediction and multimodal coordination	Misleading
NHS Greater Manchester	Patient triage and workflow optimization	Misinterpretation of unrelated AI tool

Source: Futurism (June 2026) and Financial Times reporting.

Lessons for AI Adoption in Consulting Firms

The case underscores that human verification remains essential even when AI tools are used to draft reports. As consulting firms accelerate AI agent deployment for content creation, the absence of robust fact-checking mechanisms creates downstream risks. Organizations referencing third-party reports should cross-verify specific claims against multiple independent sources, especially in emerging domains such as agentic AI where concrete implementation details are often sparse.

Frequently Asked Questions

Q: What specific claims did the KPMG report make?

It listed four detailed AI agent use cases at major clients and positioned them as evidence of successful agentic AI adoption.

Q: Which claims were disproven by the Financial Times?

All four examples were found to be inaccurate, misleading, or based on misinterpretation of unrelated announcements.

Q: What does “second

hand hallucinations” mean in this context? A: It refers to AI-generated falsehoods that enter authoritative documents and then spread further through citations by other media and organizations.

Q: How did KPMG respond to the verification?

The firm removed the report from its websites after notification, but citations had already occurred.

Q: What should companies learn from this incident?

Any AI-generated content used in reports or decision-making must undergo rigorous, multi-source human verification before publication or reliance.

Q: How can readers assess the reliability of similar consulting reports?

Cross-check concrete examples directly with the named organizations and compare against independent third-party reporting rather than accepting the report at face value.

Conclusion

The KPMG AI agent report incident illustrates the information-quality risks that accompany rapid AI adoption. The Financial Times verification provides a clear reminder that even prestigious consulting firms require stronger safeguards when incorporating AI-generated material. For readers applying AI in their own work, establishing habits of multi-source verification will become increasingly important as agentic systems proliferate.

Related articles:

Author

krona23

Over 20 years in the IT industry, serving as Division Head and CTO at multiple companies running large-scale web services in Japan. Experienced across Windows, iOS, Android, and web development. Currently focused on AI-native transformation. At DevGENT, sharing practical guides on AI code editors, automation tools, and LLMs in three languages.

DevGENT about →

📚 Read Next