AI agents in audit: more coverage, new blind spots

Описание: Article link: https://open.substack.com/pub/johanos...

This article explores how AI agents are changing the audit profession by widening what auditors can test. Instead of relying mainly on sampling, agentic systems can ingest large client data sets, reconcile accounts, flag anomalies, and suggest where teams should focus next. Done well, that means broader coverage, faster identification of risk, and fewer manual errors — a genuine step forward for audit quality.

But the article also warns about new blind spots: over-trust in machine outputs, hidden assumptions in the data, and the risk that juniors lose the apprenticeship that builds real judgement and professional scepticism. The central message is that coverage is not the same as assurance. Audit firms will need stronger governance, clearer AI audit trails, and redesigned training pathways so that humans remain accountable and scepticism becomes sharper, not softer.