Protected

Protection means keeping data, users and systems safe. This includes protecting people’s privacy, reducing cyber risks, and making sure that AI is not misused or harmful to people or organisations.

How do we use this principle?

• We will design AI systems and use cases with privacy and security from the start by embedding privacy and security design practices throughout the AI lifecycle.

• We will limit access to AI systems and data by using strong access controls, role-based permissions, and clear audit trails to prevent unauthorised use of AI.

• We will train team members on AI-specific safety and privacy risks by providing access to useful materials and support.

• We will monitor AI systems and use cases for emerging risks after deployment by establishing alerting, logging, and response protocols to catch harmful behaviour, drift or breaches.

• We will detect, prevent, and address the misuse of AI systems by identifying how AI might be used poorly or maliciously, and building guardrails to reduce these risks.