KPMG, one of the world's largest consulting firms, published a report intended to convince business clients to adopt artificial intelligence — and filled it with case studies that were fabricated, according to The Decoder.
The invented examples referenced real, well-known organizations including UBS and the NHS, lending the false claims an air of credibility. The problem was uncovered in part by Edward Tian, CEO of AI-detection company GPTZero.
Tian has a name for what happened here: "secondary hallucinations." The term describes a pattern where AI-generated misinformation doesn't just stay in a chatbot — it gets picked up, published, and amplified by authoritative institutions like consulting firms, giving bad information a veneer of trustworthiness that makes it far harder to question or correct.
The episode is particularly striking because KPMG was essentially using a document riddled with AI errors to argue that businesses should trust and invest in AI. The firm has since acknowledged the issues in the report.
This matters because when a Big Four consulting firm puts its name on a claim, executives and board members treat it as reliable intelligence. If those claims were generated or distorted by AI tools and never properly fact-checked, flawed data can quietly shape billion-dollar business decisions — and no one thinks to verify the source.