Model Extraction Attack
A type of attack that aims to reconstruct a target model's parameters, architecture or decision boundaries through systematic querying of the model's API.
In plain language
A sophisticated attack where an adversary repeatedly queries an AI to reverse-engineer its logic and create a clone. More advanced than simple model theft.
Why this matters
Model extraction threatens intellectual property and can reveal proprietary business logic. Your risk assessment should evaluate this threat for any model exposed via APIs.
Relevance
GovernanceAssessing and controlling model extraction risk is part of comprehensive AI security governance, particularly for proprietary or commercially sensitive systems.
Related terms
Putting model extraction attack 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.