EY adds AI agents to 160,000 audits
EY puts AI agents into all steps of audit work, but it gave no test results on quality.
EY started a worldwide roll-out of AI agents in its audit work. The agents are software that does many tasks without help. They work in EY Canvas, the one audit platform for 160,000 audits in more than 150 countries. EY reports that 130,000 staff use it. By 2028, EY reports, the agents will be in all audit work, from start to end.
How it works
EY put a multi-agent framework into EY Canvas. This is a group of AI agents that do tasks together. It uses Microsoft Azure, Microsoft Foundry and Microsoft Fabric. EY reports that the agents help audit staff to find risks and to use audit and accounting guidance that is always new.
The numbers
EY Canvas has more than 1.4 trillion lines of journal entry data each year. EY names the roll-out part of a multibillion-dollar commitment to audit quality. It tried the agents in tests and pilots first, but it gave no results.
What is not known
EY gave no data on errors found, time or cost. It reports that human judgment and skepticism stay. It does not tell how a person checks the work of the agents.
Insight: Auditors want agents to do the steps and a person to own the result
EY reports that human judgment stays in its audits. In November 2026 the Hong Kong securities regulator told that a person cannot check each middle step of AI agents. The regulator will keep a person in charge of the full workflow. This reads as a rule of work: the agent does the steps, and one person is in charge of the result. Clients can tell EY to name that person.
Sources
Posted