EY Deploys Autonomous AI for Audits, Forcing a Profession to Adapt

Ernst & Young is placing a major strategic bet on autonomous software to perform core audit work. The firm has introduced a new system, EY.ai Agentic Assurance, designed to have AI agents execute tasks like transaction testing and anomaly detection—work traditionally done by junior accountants. This isn't a simple software add-on; it's an architectural shift for a service governed by strict legal and regulatory standards.

The move signals a pivot. While the Big Four have long used AI in advisory roles, applying it directly to the audit—a function built on liability and public trust—is a different proposition. EY's framework operates on 'supervised autonomy,' where agents work within set boundaries but human auditors retain final approval. The impetus is clear: a severe shortage of accounting graduates and ever-more-complex financial reporting standards are straining the traditional labor-intensive model.

Financially, the potential is significant. AI can analyze entire datasets, not just samples, making 100% transaction testing a practical reality. This could alter the fundamental value and quality expectations of an audit. However, substantial hurdles remain. Regulatory bodies have yet to establish clear guidelines for evaluating AI-generated audit evidence. The reliability of large language models and questions about data security in these systems present real risks that firms must navigate.

EY's competitors—Deloitte, PwC, and KPMG—all have advanced AI initiatives, but EY's branded, audit-specific framework raises the stakes. For smaller firms, the cost of such technology could create a formidable gap. The profession now faces a dual challenge: integrating these powerful tools while ensuring they uphold, rather than erode, the professional skepticism that is the audit's cornerstone. EY's experiment will be a case study for the entire industry.

Source: Webpronews