KPMG's AI Consulting Tools Cited for Hallucinations in New Investigation

GPTZero, the AI detection firm, published an investigation Wednesday alleging that KPMG's AI-powered consulting tools have produced hallucinations — fabricated or factually wrong outputs — while the firm markets these systems as capable of "redefining excellence" for enterprise clients. The report includes documented examples of AI-generated outputs from KPMG engagements containing errors that were reportedly presented to clients without adequate verification.

KPMG has invested heavily in AI across its advisory, audit, and tax practices, partnering with Microsoft to deploy Copilot-based tools at scale. The firm has positioned AI as central to its competitive strategy, pitching clients on faster turnaround and enhanced analytical depth. GPTZero's findings, if accurate, expose a gap between that pitch and the quality controls actually in place.

The implications extend well beyond one firm. The Big Four collectively serve as financial gatekeepers for the global economy — their audit opinions carry legal weight, their advisory work shapes corporate strategy at Fortune 500 companies. An industry built on professional skepticism and documented evidence faces a credibility test when its own AI tools cannot be trusted to get basic facts right. Regulators who have already scrutinized AI use in financial services are unlikely to let this pass quietly.

Read the full story at GPTZero