KPMG Deployed 100+ AI Agents in a Real Audit and Found They Asked Better Questions Than Humans
The Decoder · rss · 2026-11-18
KPMG's "Project Exponential" deployed over 100 AI agents alongside the regular audit of a UK-listed company.
- UK head of audit Cath Burnet said the agents asked "better questions" than the human audit team in some cases, digging deeper on individual issues.
- Unlike traditional sampling, agents can analyze far more data, potentially surfacing many more findings.
- The UK accounting regulator FRC sees a risk: AI-uncovered findings could serve as a sales pitch for lucrative consulting work and undermine auditor independence.
More from AGI Musings
- Friend creator Avi Schiffmann rants against screens, calls for trials of Meta and Apple — ctjlewis · 2026-11-18
- Maryland bar counsel takes lawyer to court over ChatGPT hallucinated law, a state first — AIDidWhatJames · 2026-11-18
- Capability Is Not Authority: The Hidden Risks of Deploying AI Agents — Adjust-the-rhythm · 2026-11-18
- Claude already contributes to ~25% of Anthropic's R&D amid accelerating AI race — emmanuelvivier · 2026-11-18
- Dan Faggella: Even Pausing AGI Won't Stop Radical Change From Current AI — danfaggella · 2026-11-18
- Reddit debate: If AI may take our jobs, why not build a no-work society instead? — say-what-floris · 2026-11-18