KMPG now wants to investigate the case internally.
The management consultancy KPMG was caught giving artificial intelligence (AI) too much freedom in a report on AI case studies. Specifically, it is about the text Total Experience: Redefining Excellence in the Age of Agentic AI.
The consultants were convicted by the research group GPTZero, as the Financial Times reports.
Nach Vibe Coding nun Vibe Citing
The report published by KPMG in October 2025 described, among other things, how the Swiss bank UBS, the British healthcare organization NHS, the transport authority Transport for London and the Swiss Federal Railways use AI agents in their operations.
But that wasn’t the case. All affected organizations have now stated that the information published is inaccurate. KPMG subsequently retracted the report and launched an internal investigation. In total, only five of the 45 sources given are said to have been correct.
Not the first consultants to cheat
Shortly before, GPTZero had already caught up with colleagues from Ernst & Young in a similar way: As the company reports, EY Canada had published a 44-page report on cybersecurity at the end of 2025 entitled Points of Attack: Uncovering Cyber Threats and Fraud in Loyalty Systems. Although the report is attributed to three employees, according to GPTZero, it was “a hodgepodge of fabricated sources (‘vibe citations’), misattributions, made-up statistics and obviously AI-generated text passages.” (tf)
