Trojan Attack
A type of backdoor attack where a model is trained to behave normally on standard inputs but produces malicious outputs when a specific trigger pattern is detected.
In plain language
A hidden sleeper attack embedded in an AI. The model behaves perfectly during testing, but when a specific trigger appears in real use, it switches to malicious behaviour.
Why this matters
Trojan attacks represent an advanced threat that cannot be detected through standard testing alone. Your governance framework should address model provenance and integrity verification, particularly for critical AI systems and third-party models. This is essential for high-stakes applications in regulated industries.
Relevance
GovernanceIdentifies a sophisticated attack vector that standard testing cannot catch, necessitating governance controls around model provenance, supply chain security and adversarial testing.
Related terms
Putting trojan 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.