A major KPMG report on AI was found to contain widespread false citations and AI-generated errors, raising concerns about how misinformation spreads through influential corporate research.

What happened

GPTZero researchers examined a KPMG report on agentic AI and found that of the 45 citations, only five accurately pointed to real sources. The remainder were either entirely false or significantly distorted, including wrong attributions, paraphrased titles, and cases where topics were mistaken for authors.

Why it matters

KPMG's findings carry global influence and are likely to be cited across news reports, blog posts, and other conversations. The report is also being cited in large language models (AI systems that generate text), which could spread the misinformation further. GPTZero argues that these false citations—what they call 'vibe citations'—represent a clear danger to researchers, academics, consultants, and anyone searching the internet for information.

What to watch

This follows a similar 2025 report revealing that a study from the US Presidential Commission to Make America Healthy Again also included 'garbled or fabricated' footnotes, suggesting the problem may extend beyond a single incident to a broader pattern in research credibility.