One of the most authoritative reports in the world of consulting turned out to be a product of artificial imagination. KPMG, an international auditing firm belonging to the 'Big Four' of global consultants, was forced to withdraw its large-scale study 'Total Experience: Redefining Excellence in the Age of Agentic AI'. The cause of the scandal was the analysts' complete trust in generative neural networks, which produced a massive fabrication instead of facts.

A technical investigation conducted by specialists from GPTZero, a company specializing in AI content detection, revealed shocking details. It turned out that behind the facade of professional expertise lay a chaos of fabricated facts and non-existent sources.

The 'Vibe Citing' Method: When AI Writes Sources Based on Mood

An audit of the 45 final references used in the report text showed catastrophic results for the auditors' reputation. Out of the entire list, only 5 citations actually led to real, truthful sources of information. The other 40 links were fabricated by language models.

Researchers named this phenomenon 'vibe citing' (citing by feeling). In 28 cases, the AI took real organization or article names but significantly paraphrased the headlines and attributed fabricated data to them. In another 12 cases, the sources were described so vaguely that their physical existence could not be verified.

According to expert estimates, approximately half of the information points in the report were completely made up. This happened because the automated search engine tried to execute the task too literally and began to artificially 'find' examples of autonomous AI agents where they never existed.

Corporations Deny Successes That Never Happened

The KPMG report detailed cases of large multinational corporations that allegedly implemented agentic AI into their business. However, after publication, the mentioned companies were forced to issue official denials:

  • Emirates: Consultants claimed that the airline launched a smart chatbot named Sara, which independently changes ticket bookings. In reality, Sara is a mobile assistant from the 2023 model, lacking AI functionality and the rights to change bookings.
  • UBS: The Swiss bank denied information about the integration of AI agents into investment consulting and risk management, calling these data 'factually incorrect'.
  • SBB: The press office of the Swiss Federal Railways completely refuted the thesis that their passengers plan trips using AI carbon footprint optimizers.

Chain Reaction: How a Hoax Poisoned the Global Internet

GPTZero CEO Edward Tian warned about the 'extremely dangerous' consequences of this precedent. Since 'Big Four' reports are traditionally considered a benchmark of reliability, they are actively cited by other researchers and media.

The situation is exacerbated by the fact that the false theses from the withdrawn KPMG report have already made it into the indexing algorithms of ChatGPT by OpenAI and Gemini by Google. Now, these models are presenting the consultants' fabricated cases as real facts, creating a dangerous chain reaction of secondary AI hallucinations at the global internet level.

Currently, KPMG is conducting an internal investigation regarding the irresponsible use of AI tools in its work.