KPMG Caught Using AI to Write Report Praising AI—And It Made Up Almost Every Fact
The corporate world's obsession with artificial intelligence has reached its ultimate, peak comedy form. When one of the biggest consulting giants in the world decides to lecture everyone on the magic of automation, you expect polished slides—not a total self-own.
The auditing giant KPMG quietly pulled its glowing report on "agentic AI" after an investigation by GPTZero revealed the document was a massive, hallucinated mess. It turns out the consulting firm let an AI write the very text that was supposed to convince global executives to buy more AI consulting services.
Of the 45 citations proudly listed in the report, only five actually led to real, correct sources. The remaining 40 links were a chaotic mix of broken URLs, distorted references, and completely fictional web pages that existed only in the neural networks of a hyperactive chatbot.
The Financial Times decided to fact-check the case studies and contacted the major institutions featured in the report. Representatives from giant wealth manager UBS, the British NHS, Swiss Federal Railways, and Transport for London all confirmed that the descriptions of their AI achievements were either wildly inaccurate or completely fabricated.
Some of the hallucinations were spectacularly lazy. To prove that the Swiss railway network was ahead of the curve, the report cited a press release from 2019—a full five years before the term "AI agents" even entered the tech lexicon.
The automated writer also struggled with basic reading comprehension, repeatedly confusing the journalists who wrote the articles with the corporate subjects of the stories.
Most embarrassingly, the AI actively contradicted KPMG's own official data. The report claimed that 55% of CEOs prioritize AI investments, while KPMG's very own "CEO Outlook" published in the exact same month stated the actual number was 71%.
The spectacle of a multi-billion-dollar auditing firm failing to audit its own automated marketing brochure is the perfect summary of the current tech bubble. After getting caught red-handed, the company took down the report and issued a statement reminding staff about "human oversight"—proving that the best way to sell AI is to hope nobody actually reads what it writes.
Source: The Register
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.