Event Summary
This half-day program will address how internal audit provides assurance over AI the organization uses as well as how audit teams can responsibly use AI in their own work.
The first part of the session focuses on auditing of AI. As AI moves from pilots into core business processes, participants examine what effective AI governance looks like in practice, including the controls, oversight, and accountability mechanisms that support responsible use while managing regulatory, operational, and reputational risk. Topics include evaluating governance frameworks, model risk management, and gaps in data quality, bias monitoring, and transparency.
The second part turns to auditing with AI. Through discussion and a live demonstration/workshop, participants explore how to design and govern AI agents that support control testing without compromising professional accountability. Topics include separating agent-drafted work from professional judgment, anchoring output to authoritative guidance (COSO 2013, PCAOB, AICPA, SEC, and IIA), preserving evidence integrity, and writing prompts as governed testing methodology.
Participants leave with a practical approach for answering the board’s AI questions and a framework for building AI-assisted testing workflows that are efficient, consistent, and defensible.This half-day program will address how internal audit provides assurance over AI the organization uses as well as how audit teams can responsibly use AI in their own work.
The first part of the session focuses on auditing of AI. As AI moves from pilots into core business processes, participants examine what effective AI governance looks like in practice, including the controls, oversight, and accountability mechanisms that support responsible use while managing regulatory, operational, and reputational risk. Topics include evaluating governance frameworks, model risk management, and gaps in data quality, bias monitoring, and transparency.
The second part turns to auditing with AI. Through discussion and a live demonstration/workshop, participants explore how to design and govern AI agents that support control testing without compromising professional accountability. Topics include separating agent-drafted work from professional judgment, anchoring output to authoritative guidance (COSO 2013, PCAOB, AICPA, SEC, and IIA), preserving evidence integrity, and writing prompts as governed testing methodology.
Participants leave with a practical approach for answering the board’s AI questions and a framework for building AI-assisted testing workflows that are efficient, consistent, and defensible.
Learning Outcomes
Upon completion of this program, participants will be able to:
Auditing of AI
- Evaluate AI governance frameworks and the controls, oversight, and accountability mechanisms that support responsible AI use.
- Explain model risk management principles as they apply to AI systems moving from pilot to production.
- Identify gaps in data quality, bias monitoring, and transparency that create regulatory, operational, or reputational risk.
- Develop an approach for responding to board and audit committee questions about AI, whether internal audit is advising on AI projects, reviewing existing controls, or both.
Auditing with AI
- Determine which audit judgments must remain with credentialed professionals when using AI agents, including sample selection, population completeness, report reliability, deficiency classification, and sign-off.
- Describe governance controls for AI-assisted testing, including confidence labeling, evidence preservation, and escalation of potential fraud, management override, or illegal acts to a human reviewer.
- Apply prompt design techniques such as role definition, methodology-based instructions, closed result vocabularies, standardized intake, and explicit prohibitions to produce consistent, reviewable output.
- Evaluate prompt effectiveness by testing against known clean samples and exceptions, and establish ownership, version control, and change management for prompts used in production.