KPMG Retracts AI Research Following Widespread Data Hallucinations
KPMG withdraws AI research after widespread data hallucinations raise concerns about AI accuracy, reliability, and enterprise trust.
KPMG has officially withdrawn its flagship research report, “Redefining Excellence in the Age of Agentic AI,” after an investigation revealed that the document contained significant fabricated content. The retraction, confirmed on June 13, 2026, followed reports that the study’s central case studies were largely products of AI hallucinations rather than verified corporate data.
The inaccuracies were first identified by the AI-detection firm GPTZero and subsequently verified by the Financial Times. Investigations revealed that out of 45 citations in the report, only five were accurate. The report incorrectly attributed advanced AI agent deployments to several high-profile institutions, including UBS, the UK’s National Health Service (NHS), Swiss Federal Railways, and Transport for London. Each of these organizations formally contested the claims, describing them as either “factually incorrect,” “misleading,” or entirely inconsistent with their actual technological operations.
This incident highlights a growing concern about “vibe citing”, where generative AI creates plausible but entirely false references and case studies. For the B2B sector, this represents a significant reputational risk. As large consulting firms increasingly market themselves as experts in AI adoption, the failure to apply human oversight to their own research pipelines creates a dangerous ripple effect. Because the report was cited by multiple technology publications and European news outlets before its removal, it has contributed to the spread of misinformation across the digital ecosystem.
The KPMG retraction arrives amid a documented decline in consumer trust regarding AI-generated media. According to recent survey data from Gartner, 49% of U.S. consumers and 57% of Millennials and Gen Z believe that the proliferation of generative AI has actively worsened the quality of online content. As enterprise buyers become increasingly skeptical of automated outputs, the industry is seeing a sharp move away from fully autonomous content pipelines. Moving forward, B2B firms are expected to shift toward hybrid verification models that require human subject-matter experts to validate every data point before publication.
