What is the difference between audit AI and general-purpose AI?
General-purpose AI is designed to help with a very wide range of tasks. Audit AI is configured around specific audit workflows, evidence requirements, methodologies and review controls.
A general model may write a convincing summary, but it does not automatically know what constitutes sufficient support for an audit conclusion, which risk framework the firm applies, how contradictory evidence should be handled or what needs to be retained on the file. Those controls have to be built into the product and the firm's use of it.
Purpose-built audit AI should provide:
- a defined workflow rather than an open-ended chat;
- controlled access to engagement evidence;
- conclusions linked to specific source material;
- firm-configured risk and control language;
- visible review, override and approval steps;
- outputs designed for the audit file;
- appropriate security, retention and access controls.
General-purpose AI can still help auditors with low-risk productivity tasks. The closer the task gets to audit evidence or a documented conclusion, the stronger the case for a purpose-built and properly governed system.
References: FRC AI in Audit · Platformed security and audit controls