ELSEC AI

Human Agency and Oversight

The requirement relates to the capability of users to make informed and autonomous decisions, the right not to be subject to a decision based solely on automated processing for important matters, the capability to oversee the overall activity of the AI system (including its broader economic, societal, legal and ethical impact); and the establishment of adequate levels of human discretion during to use AI systems.

Technical robustness and safety 

The requirement relates to the development of AI systems that are protected against vulnerabilities (such as hacking), guaranteeing the systems make correct predictions, recommendations, or decisions, ensuring that systems work properly under a range of circumstances and situations, and ensuring systems are reproducible and provide consistent results.

Privacy and data governance

The requirement relates to the establishment of guarantees so that AI systems ensure privacy and data protection, ensure the quality of the data used, and the implementation of protocols to govern the access to data.

Transparency

The requirement relates to the establishment of traceability mechanisms to document the development of the system (including data and algorithms), the ability to explain both the technical processes and the human decisions, the right to be informed when interacting with an AI system, and the right to be informed appropriately about AI system’s capabilities and limitations.

Diversity, non-discrimination and fairness

The requirement relates to the establishment of adequate mechanisms for AI systems to avoid unfair bias that could lead to prejudice and discrimination; the development of AI systems adequate for people regardless of their age, gender, abilities and other characteristics; and the consultation of stakeholders who may directly or indirectly be affected by the system. 

Societal and environmental wellbeing

The requirement relates to the establishment of measures to secure the environmental friendliness of AI systems’ entire supply chain the assessment and monitoring of the effects of AI systems on individuals,  society and democracy.

Accountability

The requirement relates to the establishment of procedures to audit AI systems, the implementation of procedures to minimise and report negative impacts, the reporting of decisions related to trade-offs, and the definition of mechanisms to seek reparations for damages or prejudices linked to the use of AI systems.