1.BV | Fairness in Build and Validate: fairness-conscious feature choices, champion model within fairness thresholds, causal feature selection and safeguards against discriminatory deploy | 0 |
1.DM | Fairness in Deploy and Monitor: measure outcome fairness, monitor metrics against thresholds and quantify harm and benefit for sensitive groups | 0 |
1.PD | Fairness in Plan and Design: fairness-aware design, harm and benefit analysis, chosen fairness metrics and defined sensitive attributes with thresholds | 0 |
1.PID | Fairness in Prepare Input Data: representative, accurate data of sufficient size, with sensitive attributes and their proxies excluded or justified | 0 |
2.BV | Privacy and Security in Build and Validate: privacy and security by design, protected model architecture, integrity under adversarial conditions, and testing that combined data doe | 0 |
2.DM | Privacy and Security in Deploy and Monitor: continuous monitoring, revisited privacy and risk assessments, and owner accountability for lifelong protection | 0 |
2.PD | Privacy and Security in Plan and Design: privacy-respecting design, minimal personal data, approved profiling only, security blueprint, applicable standards, data classification an | 0 |
2.PID | Privacy and Security in Prepare Input Data: lawful data handling, confidentiality and delegated authority, integrity and availability, contextual classification with de-identificat | 0 |
3.BV | Humanity in Build and Validate: decision-making aligned with human rights and cultural values, with parameters and training that advance humanity | 0 |
3.DM | Humanity in Deploy and Monitor: periodic assessment against human rights, cultural values and KPIs with corrective measures | 0 |
3.PD | Humanity in Plan and Design: ground the model in fundamental human rights and KSA cultural values and define how it augments human capability | 0 |
3.PID | Humanity in Prepare Input Data: responsible and ethical data management under KSA data regulations | 0 |
4.BV | Social and Environmental Benefits in Build and Validate: tie results to a recognised social purpose and keep resource and energy use acceptable | 0 |
4.DM | Social and Environmental Benefits in Deploy and Monitor: continuous assessment of human, social, cultural, economic and environmental impact against design objectives | 0 |
4.PD | Social and Environmental Benefits in Plan and Design: design to prevent harm to people and the environment and to address social and ecological issues | 0 |
4.PID | Social and Environmental Benefits in Prepare Input Data: follow data management policies and make social and environmental data publicly accessible with its benefit stated | 0 |
5.BV | Reliability and Safety in Build and Validate: comprehensive testing methodology, rigorous validation with oversight and integration sign-off, human determination for irreversible d | 0 |
5.DM | Reliability and Safety in Deploy and Monitor: continuous monitoring of robustness inward and outward, conformity with the designed structure, and prevention of destructive use | 0 |
5.PD | Reliability and Safety in Plan and Design: design for uncertainty, robust operation across inputs, reliability standards, documentation and stakeholder sign-off at critical decisio | 0 |
5.PID | Reliability and Safety in Prepare Input Data: measure data quality, accuracy, suitability and credibility, and prepare stress test data for extreme scenarios | 0 |
6.BV | Transparency and Explainability in Build and Validate: transparent algorithms in process and product, explanatory information on request, and AI ethics due diligence on third-party | 0 |
6.DM | Transparency and Explainability in Deploy and Monitor: publish performance and impact metrics, log and disclose failures and breaches, and test UI and UX periodically | 0 |
6.PD | Transparency and Explainability in Plan and Design: inform affected stakeholders, give access to understandable rationale, set transparency levels, build an information section and | 0 |
6.PID | Transparency and Explainability in Prepare Input Data: document datasets and decision processes for traceability, and assess data accuracy, suitability, validity and source | 0 |
7.BV | Accountability and Responsibility in Build and Validate: owned modelling decisions, assigned responsibilities, quantitative and qualitative justification, and sign-off after valida | 0 |
7.DM | Accountability and Responsibility in Deploy and Monitor: clear responsibility and liability, continuous monitoring with periodic performance reports, and predefined triggers assign | 0 |
7.PD | Accountability and Responsibility in Plan and Design: attributable responsibility, a governance structure without gaps, supporting instruments and human-controlled design | 0 |
7.PID | Accountability and Responsibility in Prepare Input Data: data quality checks, approved and signed-off datasets cleansed of societal bias, and documented, validated preparation | 0 |
annex | Annexed AI ethics tools, their lifecycle mapping and the AI Ethics Checklist | 0 |
intro | Purpose, scope, definitions and the AI System Lifecycle | 0 |
risk | AI risk tiers and risk management embedded in the lifecycle | 0 |
roles | Adopting entities designate the Head of Entity or CDO, a Chief Compliance Officer, a Responsible AI Officer and an AI System Assessor with defined duties, and bind third-party syst | 0 |
sdaia | SDAIA's national role, optional registration, compliance measurement and motivational badges | 0 |