Machine Unlearning
The technical process of removing the influence of specific training data from a trained machine learning model, enabling compliance with data deletion requests and privacy regulations.
In plain language
The technical process of making an AI "forget" specific data it was trained on. If someone exercises their right to be forgotten under privacy law, machine unlearning removes the impact of their personal data from the model without retraining the entire system.
Why this matters
Machine unlearning is an emerging governance requirement driven by privacy rights and regulatory expectations. As the Privacy Act's Australian Privacy Principles evolve and as the Voluntary AI Safety Standard develops, your organisation should track developments in unlearning feasibility and prepare strategies for managing data deletion requests in relation to deployed AI systems.
Relevance
GovernanceMachine unlearning addresses privacy rights by enabling AI systems to forget specific individuals' data without full retraining.
Related terms
Putting machine unlearning into practice in your organisation?
Ready to transform your AI strategy?
Partner with Australia's AI strategy and governance specialists. From adoption roadmaps to ISO 42001 audit readiness.