Newsletter: AI in IFRS Compliance — From Automation to Governed Intelligence

Prepared by Hafsa Research and Analysis Company

The Core Shift: AI Is Now a Governance Object, Not Just a Tool

AI in IFRS compliance has moved past the experimentation phase. The question is no longer whether AI improves efficiency—the evidence is clear. AI adoption shows a statistically significant positive relationship with audit quality (β = 0.347, p < 0.01), with the model explaining approximately 61.8% of variation in audit quality among Nigerian audit firms . For fraud detection, AI demonstrates an even stronger effect (β = 0.42, p = 0.0186, R² = 0.79) .

But the conversation has shifted. Two landmark publications in early 2026 changed the governance landscape: COSO released its first guidance on generative AI internal controls (February 2026) , and the UK Financial Reporting Council (FRC) issued guidance on generative and agentic AI in audit (March 2026) .

The FRC’s message is unambiguous: “While technology changes, the fundamental principle of our regulatory framework does not: it is people—the firms and Responsible Individuals—who are accountable for audit quality” .

Solution 1: Establish Cross-Functional AI Governance with a Use-Case Inventory

COSO’s six-step roadmap—govern, inventory, assess, design, implement, monitor—provides the operational framework . The critical first step is eliminating “shadow AI”: generative AI tools deployed without formal oversight.

The guidance emphasizes that “set-and-forget” does not work with probabilistic models. Continuous monitoring is necessary to detect model drift, prompt degradation, or vendor changes .

Action step: Conduct a formal AI use-case inventory across finance, audit, and IT. Map each use case to specific IFRS processes and key controls. Identify any unapproved tools in use and either formalize or eliminate them.

Solution 2: Apply Heightened Rigor to Financially Relevant AI Outputs

Not all AI use cases carry equal risk. COSO’s guidance recommends a use-case-based decision framework to “right-size the level of human involvement” based on how directly AI outputs influence financial reporting .

For AI outputs affecting material amounts in financial statements—journal entries, reconciliations, impairment estimates, ECL calculations—the guidance requires:

• Appropriate human oversight
• Supporting evidence
• Documentation of review and approval

Action step: For each IFRS process using AI (IFRS 9 ECL modeling, IFRS 15 revenue recognition, IFRS 16 lease classification), define the human checkpoint. Document what constitutes sufficient appropriate evidence for AI-assisted conclusions.

Solution 3: Build Control “Building Blocks” for AI Across the Organization

COSO’s guidance identifies six control categories that should be implemented for every AI use case :

1. Access and acceptable-use restrictions — including vendor tools and plugins
2. Input/data controls and retrieval constraints — ensuring data quality and relevance
3. Prompt/configuration governance and change control — tracking model version changes
4. Output validation and exception handling — acceptance and accountability for outputs
5. Logging/traceability — model/version, prompts, key inputs/outputs, approvals
6. Monitoring controls — for drift, anomalies, and unauthorized use

Action step: Audit your current AI controls against these six building blocks. If any are missing for financially relevant use cases, prioritize remediation before scaling further.

Solution 4: Preserve Professional Skepticism Through Structured Challenge

Research on Big 4 auditors’ GenAI adoption identifies a critical risk: “overreliance without verification, lack of transparency, confidentiality and security risks, and inability to exercise skepticism and judgment can have an adverse impact on audit quality” .

The FRC guidance similarly emphasizes that AI tools support, but do not replace, professional judgment . The human auditor remains accountable for all conclusions.

Action step: Implement mandatory challenge protocols for AI-flagged items. Auditors and finance professionals must articulate why they accepted or rejected AI outputs—not simply confirm the system’s conclusion.

Executive Checklist: AI Governance Readiness

Governance:

• □ Establish cross-functional AI governance with defined roles and escalation protocols
• □ Maintain a formal AI use-case inventory (including shadow AI)

Controls:

• □ Map AI use cases to IFRS processes, assertions, and key controls
• □ Implement COSO’s six control building blocks for each financially relevant use case

Documentation:

• □ Define evidence standards for AI-assisted work
• □ Coordinate with internal and external auditors on documentation requirements

Skepticism:

• □ Implement structured challenge protocols for AI outputs
• □ Train staff on recognizing AI limitations (hallucination, overconfidence, drift)

Closing Thought

AI is not a threat to IFRS compliance. But ungoverned AI is. COSO’s guidance makes clear that generative AI “does not replace internal control. It must operate within it” .

The FRC’s guidance adds the regulatory dimension: “AI is a powerful tool, but the professional judgement and accountability of the auditor remains at the core” .

For CFOs and finance leaders, the path forward is not to resist AI adoption—but to build the governance architecture that ensures AI strengthens, rather than undermines, confidence in financial reporting.

Prepared by Hafsa Research and Analysis Company