KPMG UK has run an experimental parallel audit of an FTSE 250 company using over 100 AI agents across all audit phases. The project, codenamed Project Exponential, was developed over four months and recorded agent interactions for full traceability. Cath Burnet, head of audit at KPMG UK, said the agents dug deeper into individual questions and asked better questions than human auditors in some cases.
Human oversight remained critical during the trial, as an auditor had to intervene when an AI agent failed to correctly match regulations against company documents. The UK’s accounting regulator, the FRC, warns that AI agents analyzing far more data than traditional sampling could threaten auditor independence. KPMG plans to expand the project to as many as 20 larger audits, though the technology is not yet mature enough for thousands of engagements.
This trial highlights the growing role of AI in financial oversight and the need for human judgment to catch errors in automated systems. The FRC’s warning suggests regulators are closely watching how AI affects auditor independence and data sampling methods. You should watch how these tools evolve, as they could eventually change the speed and depth of financial audits for businesses and investors alike.


