Professional services firm Kpmg has removed its October 2025 report, “Redefining Excellence in the Age of Agentic AI,” following widespread accusations of factual inaccuracy. The document, which surveyed artificial intelligence adoption across major corporations, was flagged by research group GptZero for containing numerous claims that did not match reality. Several high-profile entities, including Ubs, the UK’s National Health Service, Swiss Federal Railways, and Transport for London, confirmed to the Financial Times that the report misrepresented their actual AI usage or contained misleading data. Kpmg stated it withdrew the publication from all websites while launching an internal investigation into the matter. The firm’s spokesperson emphasised that staff are expected to adhere to strict guidelines regarding responsible AI use, specifically mandating human oversight to validate content and verify independent sources before release.
This incident highlights a growing risk in the professional services sector where firms rely heavily on generative tools to produce their own strategic analysis. It raises serious questions about the integrity of industry reports when the tools used to create them lack sufficient human verification. The situation mirrors a similar event last month when EY withdrew a report on loyalty rewards programs after discovering fake footnotes and hallucinated data. As organisations increasingly delegate content creation to algorithms, the potential for self-referential errors becomes a significant liability. The case serves as a stark reminder that automated assistance cannot replace the critical need for human accountability in high-stakes business communications.
- Kpmg withdrew its AI usage report after clients confirmed the text contained false claims generated by the very technology it analysed.
- The incident underscores the necessity of human oversight when professional firms use generative AI to draft their own strategic publications.
- This event follows a similar retraction by EY, indicating a rising trend of hallucination errors in corporate AI-generated content.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
