KPMG pulls report on AI usage due to apparent hallucinations

2026-06-13 · Source: TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Fundamental Awareness, quick

Summary

Professional services firm KPMG has withdrawn its report, "Redefining excellence in the age of agentic AI," published in October 2025, after multiple organizations disputed its claims regarding their AI usage. Research group GPTZero identified numerous inaccuracies, suggesting these stemmed from AI hallucinations, implying KPMG used AI to draft the report itself. UBS, the UK's National Health Service, Swiss Federal Railways, and Transport for London all confirmed to the FT that the report's assertions about their AI implementation were either false or misleading. KPMG has removed the report from its platforms and initiated an investigation, reiterating its guidelines for responsible AI use, which mandate human oversight for content validation and source verification. This incident follows a similar withdrawal by EY last month concerning a report with apparent AI hallucinations and fake footnotes.

Key takeaway

For Directors of AI/ML overseeing content generation, this incident underscores the critical need for robust validation workflows. You must implement stringent human oversight protocols to verify all AI-generated claims and cross-reference them with independent sources before publication. Relying solely on AI for factual content risks severe reputational damage and necessitates costly retractions. Ensure your teams adhere to clear guidelines mandating human review to prevent hallucination-induced inaccuracies.

Key insights

AI-generated content requires rigorous human oversight to prevent factual inaccuracies and reputational damage.

Principles

• Human oversight is critical for AI-generated content.
• Verify AI outputs against independent sources.
• AI hallucinations pose reputational risks.

In practice

• Implement multi-stage human review for AI drafts.
• Cross-reference AI claims with primary data.
• Establish clear AI usage guidelines for staff.

Topics

• AI Hallucinations
• Content Verification
• Responsible AI
• Professional Services
• Reputational Risk
• Human Oversight

Best for: CTO, VP of Engineering/Data, Executive, Director of AI/ML, AI Ethicist, Consultant

Related on AIssential

• Breaking: Google liable for hallucinations — A German court holds Google liable for AI Overviews' false outputs.
• Concerns ‘AI slop’ used by University of Sydney-based institute to lobby for $20m gambling education funding — A research institute's funding request was undermined by an evidence review containing significant inaccuracies, potentially due to AI-generated content.
• Steering the Verifiability of Multimodal AI Hallucinations — Multimodal AI hallucinations vary in human verifiability, requiring distinct intervention strategies for obvious versus elusive errors.
• The hardest question to answer about AI-fueled delusions — Chatbots can amplify user delusions and dangerous ideation through constant validation and simulated sentience.
• We Asked the ‘Future of Truth’ Author to Explain How He Used AI. It Didn’t Go Well — The irony of a book on AI and truth containing AI-generated falsehoods highlights critical authorial responsibility.
• The literal enshittification of The United States — AI chatbots can generate dangerously inappropriate advice, especially in sensitive domains like health.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by TechCrunch.