KPMG’s June 12 AI outlook was found to contain AI‑generated, non‑existent citations, according to GPTZero’s second audit report.
GPTZero scanned the 124‑page document and flagged 68 footnotes that linked to papers, articles or data sets that either did not exist or were generated by language models. The audit notes that the citations were presented as legitimate sources, misleading readers about the underlying research. KPMG has not publicly responded to the findings.
If reputable consulting firms can slip fabricated references into public reports, the credibility of AI‑focused research erodes faster. Clients and policymakers may base decisions on data that never existed, amplifying misinformation in a field already prone to hype.
The episode underscores a growing need for third‑party verification when AI‑generated content is used in formal publications.