KPMG pulls report on AI usage due to apparent hallucinations

The more pointed detail here is not that hallucinations occurred, but that KPMG, a firm that sells AI advisory services to clients, was caught using LLMs in its own knowledge production without sufficient verification controls. That's a credibility problem with direct commercial consequences, not just a workflow embarrassment.

This is largely disconnected from recent activity in our archive, as we have no prior coverage to anchor it to. It belongs, however, to a broader pattern playing out across professional services: the gap between how firms position AI to clients and how rigorously they govern it internally. Consulting firms have been among the loudest advocates for enterprise LLM adoption, which makes this incident particularly awkward. The liability question it raises, specifically who is responsible when AI-assisted deliverables contain errors, remains genuinely unresolved across the industry and is not a problem any major vendor has addressed in published terms of service.

Watch whether KPMG publishes a revised report with an explicit methodology disclosure about AI use, and whether competing firms (Deloitte, PwC, McKinsey) respond with public QA policies of their own within the next 90 days. Silence from peers would suggest the industry is hoping this blows over rather than treating it as a forcing function for standards.