Integrating Privacy-Preserving AI Models into AI Governance Frameworks

The integration of Artificial Intelligence (AI) into Information Technology environment has transformed organizational processes, yet it has also introduced challenges around privacy, accountability, and regulatory compliance. This study explores how privacy-preserving AI (PPAI) techniques can strengthen IT governance and compliance, and identifies the governance controls internal IT auditors require in AI developments. A qualitative exploratory research design was adopted, drawing from nine peer-reviewed articles, regulatory framework (as GDPR, EU AI Act, the NIST AI RMF, and ISO/IEC standards) IT governance models (COBIT, ISACA guidelines, ISO/IEC 27001). The analytical process combined thematic analysis and comparative mapping to PPAI techniques with IT governance control and assurance checkpoints. The findings reveal that federated learning operationalizes privacy-by-design by minimizing raw data transfer, aligning with GDPR and similar principle, Secure aggregation, homomorphic encryption, and differential privacy strengthen confidentiality and safeguard model outputs against inference attacks, while immutable logging and explainability provide accountability and auditability consistent with ISO/IEC 27701 and NIST AI RMF. From an assurance perspective, auditors must expand evaluations to cover AI-specific risks, including model integrity, federated learning protocols, and privacy-preserving outputs. The study concludes that PPAI serves not only as a technical safeguard but also as a governance enabler. Recommendations include embedding PPAI in IT operations, updating governance standards, and developing dynamic audit framework tailored to AI-driven environments.