KPMG, one of the world's largest accounting and consulting firms, has pulled a report it published about the benefits of artificial intelligence after investigators found the document was riddled with AI hallucinations, according to TechCrunch and Engadget.
The report, which KPMG published last year, was apparently produced with the help of the very technology it was meant to champion. An investigation revealed that the paper contained multiple instances of AI-generated misinformation — the kind of confident, plausible-sounding falsehoods that large language models are known to produce.
TechCrunch noted the irony pointedly: "Once again, AI proves to be an unreliable source of information about AI."
The episode is embarrassing on multiple levels. KPMG is a firm whose clients pay for rigorous, trustworthy analysis. A report on AI's benefits carries implicit authority — the kind of document that shapes boardroom decisions and policy conversations. Finding out it was built on fabricated foundations undermines not just the report itself, but confidence in the due diligence behind it.
It also highlights a compounding problem in the AI discourse: companies rushing to publish pro-AI content are increasingly turning to AI to write it, creating a feedback loop where AI tools are hyping themselves — and getting the facts wrong in the process.
KPMG has not been alone in this predicament; newsrooms, law firms, and consultancies have all faced embarrassment after AI-generated content slipped through without adequate fact-checking.
This story matters because when a Big Four firm can't verify an AI-generated report about AI, it raises urgent questions about how seriously enterprises are scrutinizing the outputs of the tools they're betting their businesses on.