1 hour ago
25
KPMG’s 2025 global study on trust in artificial intelligence surveyed over 48,000 people across 47 countries. The headline number looks great for the AI cheerleaders: 73% of respondents said they’ve observed personal benefits from AI, primarily around efficiency gains. But buried in the same data is a less flattering stat. A full 56% of those surveyed reported making mistakes because they relied on unverified AI outputs.
KPMG’s own study explicitly identifies hallucinations as a major limitation of generative AI. “Hallucinations” is the industry’s term for when AI confidently presents completely fabricated information as fact.
EY had to retract a cybersecurity report in May 2026 after discovering that 16 of its 27 citations were completely fabricated by AI. Not slightly wrong. Not taken out of context. Fabricated. The sources didn’t exist.
Deloitte, meanwhile, agreed to refund portions of a government contract after AI hallucinations were found in one of its reports.
The KPMG study, conducted in collaboration with the University of Melbourne, quantifies this tension across 47 countries. Whether you’re in Tokyo or Toronto, the same pattern holds: AI delivers real productivity gains while introducing a new category of risk. Professional services firms are investing heavily in AI governance frameworks to capture AI’s efficiency benefits while building enough human oversight into the process to catch hallucinations before they reach clients.
For investors watching the AI space, the KPMG findings should serve as a reality check. The 73% benefit number explains why AI adoption continues to accelerate across industries. But the 56% mistake rate represents a significant risk in any business model that treats AI outputs as reliable without human verification.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·