KPMG pulled and withdrew a report titled Redefining excellence in the age of agentic AI after multiple organizations flagged inaccuracies and misleading claims about their use of AI agents.

Researchers describe the practice of citing benign-sounding but unverifiable sources as vibe citing, a form of AI hallucination.

GPTZero found AI hallucination signs in the report, including claims about Greater Manchester NHS predicting readmissions and routing patients that lacked substantiation from cited sources.

Experts warn that such errors could trigger a secondary hallucination effect, letting misinformation spread through media and analyses and eroding trust in AI-adjacent research from advisory firms.

GPTZero argues vibe citations have real consequences because KPMG’s influence could cause distorted findings to be cited globally across outlets.

The report is described as containing fake or heavily distorted references, including garbled attributions and paraphrased titles that misrepresent sources.

There are calls for stronger oversight of AI models, with leaders like Anthropic’s Dario Amodei advocating stricter AI regulation.

This incident follows a separate May case in which EY withdrew a loyalty rewards report that contained fake footnotes and AI hallucinations.

Concerns persist that misinformation could further spread through language models that reference the KPMG report in training or outputs.

The broader professional-services sector faces AI-related accuracy risks, with EY’s similar withdrawal showing a pattern of AI-assisted research going wrong.

GPTZero investigators found that only five of 45 citations in the KPMG report pointed to real sources, with AI-generated errors and misleading citations.

Numerous organizations, including UBS, the UK’s National Health Service, Swiss Federal Railways, and Transport for London, said the report’s claims about their AI use were false or misleading.