2.1a | Adopt an AI governance policy, a risk management procedure and data governance and protection policies | 0 |
2.1b | Adopt AI development and engineering standards and human oversight and operational controls | 0 |
2.1c | Adopt procurement and vendor governance, a transparency and user communication policy, an incident response procedure and an audit and assurance procedure | 0 |
2.2 | Build workforce readiness across leadership, risk and audit competence, data and technical skills, oversight and continuous capacity building | 0 |
3.1a | Design and development, define and prepare: purpose, requirements, stakeholders and impact | 0 |
3.1b | Design and development, design and build: justified model selection, data readiness, Frugal AI and prototyping | 0 |
3.1c | Design and development, embed trustworthiness: fairness, transparency, privacy by design, security and robustness, and the level of human oversight | 0 |
3.1d | Design and development, iterate and ensure: validate against requirements through internal audit and refine continuously | 0 |
3.2a | Pre-deployment TEVV: a formal plan aligned to standards and regulation, with independent review | 0 |
3.2b | Pre-deployment TEVV: validate performance, robustness, security and safety, fairness and ethics, and data quality and privacy | 0 |
3.2c | Pre-deployment TEVV: validate human oversight and safeguards, keep traceable documentation and obtain formal sign-off with no high or critical risk open | 0 |
3.3 | Post-deployment: monitor performance, risk and trustworthiness, operational health, audit for compliance, and improve and re-validate | 0 |
4 | Engage stakeholders: identify and engage them, collaborate, keep two-way communication with communities, communicate transparently and support responsible disclosure | 0 |
5.1 | Protect society: purposeful adoption, protection of minors, cultural alignment, user rights, human-centred decisions, responsible use boundaries and privacy-conscious behaviour | 0 |
5.2 | Protect the environment: right-size solutions, adopt Frugal AI practices, train efficiently, minimise data, assess environmental impact, choose sustainable infrastructure, monitor | 0 |
A | Annexes: the Responsible AI Evaluation Tool, trustworthy AI tools and international standards | 0 |
I | Scope, applicability, objectives and key concepts | 0 |
III | Measure readiness: the Enterprise Readiness Level and the Responsible AI System Readiness Level with their scoring bands | 0 |