KPMG fabricated AI case studies including NHS and UBS

KPMG published a client-facing report on AI in business that included multiple fabricated case studies citing organisations such as UBS and the NHS, an independent review of the document found. The examples were presented as real-world deployments to persuade clients of the business case for rapid AI adoption.

The fabrications were flagged publicly after independent sleuthing identified discrepancies between the report's claims and any verifiable public statements or project records from the named organisations. Edward Tian, CEO of GPTZero, highlighted several instances where the report attributed outcomes and partnerships to institutions that said they had not taken part in the described work.

What the report contained

The report framed a series of short case studies as evidence that large institutions were already deploying generative AI at scale to cut costs and speed decisions. Examples included a banking client named as UBS and a health service identified as the NHS. In the flagged passages the report attributed specific operational savings, pilot outcomes, or vendor relationships to those organisations.

Reviewers who examined the report found mismatches between the report's descriptions and public statements, press releases, regulatory filings, and other documentary evidence. In several instances the language in the report implied an implementation or commercial relationship that the named organisation had not acknowledged. The use of client names in marketing and advisory material is common in consulting, but those references normally rest on explicit permission or verifiable public case studies.

The document appears to have been distributed to prospective clients and used as part of KPMG's sales conversations around AI strategy and adoption. The core allegation is not that the report misstated technical capabilities, but that it attributed concrete outcomes and partnerships to institutions that did not corroborate them.

Industry reaction and next steps

The disclosure has prompted immediate criticism from privacy advocates, client-side procurement teams, and ethics observers who said the episode raises questions about how consultancy firms substantiate claims when selling advisory services. Some corporate counsel teams told contacts they would scrutinise references and examples more closely in future vendor engagements.

KPMG's public communications around the incident have been limited. External commentators noted that the episode sits at the intersection of marketing practice and professional standards for auditors and consultants, and that the potential consequences range from reputational damage to contractual disputes with clients who relied on the report in procurement decisions.

Regulators that oversee professional services and advertising may take an interest if clients can show demonstrable harm from relying on inaccurate case studies. Professional associations and large enterprise buyers often require documented consent before a supplier uses a customer name or outcome in sales materials. The contested material in the KPMG report raises whether those norms were followed.

Why it matters

The episode highlights the risks when consultancies use unverified or overstated examples to persuade clients about AI's promise. For enterprise buyers, it underscores the need for independent verification of vendor claims before committing to expensive AI projects. For regulators and professional bodies, the case amplifies pressure to clarify what constitutes acceptable evidence in advisory marketing and when misstatements cross into misconduct.