Safety Evaluation
The comprehensive testing and assessment of AI systems for potential harms, including harmful outputs, vulnerability to misuse and unintended negative consequences.
In plain language
Comprehensive testing of an AI for all the ways it could cause harm: generating dangerous instructions, producing biased outputs, being vulnerable to manipulation and so on.
Why this matters
Safety evaluation is a governance requirement before any AI system is deployed. Your framework should define safety evaluation criteria, testing methodologies and acceptance thresholds for all AI applications, with documentation proportionate to risk.
Relevance
GovernanceSafety evaluation methodologies and results must be documented and approved before deployment, with criteria tailored to the specific harms each application could cause.
Related terms
Putting safety evaluation 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.