Data Poisoning
An attack that corrupts a machine learning model by injecting malicious data into the training dataset, causing the model to learn incorrect patterns or exhibit targeted misbehaviour.
In plain language
Sabotaging an AI by sneaking bad data into its training set. Like slipping wrong answers into a student's textbook; the AI learns incorrect things and makes mistakes later.
Why this matters
Data poisoning is a significant AI security risk that your governance framework must address through data integrity controls. Supply chain verification, access controls for training data and ongoing monitoring are essential governance measures to protect against this threat.
Relevance
GovernanceData poisoning represents a supply chain and security risk that requires governance controls over data access, provenance verification and detection mechanisms.
Related terms
Putting data poisoning 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.