ST4S Framework v2024.1

Incorporating major research and development by ESA and members of the ST4S Working Group, the AI module is designed to align with the key principles of the Ministerial Framework for AI in Education.

Overall, companies will need to demonstrate a commitment to safety and security, respect privacy, and be transparent on their usage of AI.

Key features of the AI Module:

• An exclusion list applies:
  • Not all AI services are assessable and there is a focus on educational specific use cases of AI. Examples of services not assessable include using AI for biometrics, sensitive/personal information or student monitoring and administrative decision making (e.g. automated enrollment ratings with AI or analysing behaviour of students with AI).
  • Some items on the exclusion list are subject to change pending advice from government and industry.
• Privacy controls:
  • All services must opt-out information and user data or assets from training by default.
  • If a service may use information for training or further developing the AI module, the service must provide privacy controls that allow users to easily opt-in, opt-out, and customise their preferences regarding how their information and data is to be used.
• Ethical framework and appointing an ethics/safety officer:
  • Organisations must establish an ethical AI framework and appoint an AI ethics lead that is independent of development. The AI ethics lead should handle complaints from users, oversee feature development, and ensure safety testing and privacy practices are in place.
• Testing requirements:
  • Testing is required for security (e.g. jailbreaking attempts), privacy (preventing personal information being disclosed to the model), and safety (e.g. testing for inappropriate content generation and outputs).