France

The ongoing work on sustainable AI is preparing organizations for compliance with forthcoming EU regulations on environmental sustainability.
CEN CENELEC JTC21 AI WG4 Foundational and societal aspects
ISO/IEC JTC1 SC 42 AI WG3 Trustworthiness
AFNOR ethics committee on AI

The development of the AI Trustworthiness Framework is highly significant as it directly supports the implementation of the EU AI Act. This framework establishes essential standards that will enable organisations to meet the legal requirements of the Act. Furthermore, the ongoing work on sustainable AI is preparing organizations for compliance with forthcoming EU regulations on environmental sustainability. These initiatives focus on creating AI systems that are energy-efficient and environmentally responsible, ensuring that businesses are not only able to meet the new regulatory standards.

The different targeted standards have a different societal impact:
AI Trustworthiness Framework (prEN 18229): (Part 1 and Part 2) Establishes terminology, concepts, and requirements for AI trustworthiness, addressing five of the ten SRs. Facilitates AI Act compliance and meets varied stakeholder needs.
Environmentally Sustainable AI (JT021010): Cuts AI energy consumption—particularly in neural networks—through more efficient algorithms and pre-trained models, in line with EU climate neutrality targets.
Transparency Taxonomy of AI Systems (JT021022): Creates a structured framework to enhance transparency, accountability, and comparability across AI systems.
Upskilling on AI Ethics (JT021033) & Ethical Management Guidelines (JT021034): Provide tools and guidance to embed ethical and social considerations throughout the AI lifecycle.
Sustainable AI – Guidelines and Metrics (JT021035): Defines KPIs to assess and minimise AI’s environmental impact, promoting sustainable practices.
Impact Assessment and Fundamental Rights (JT021026): Identifies and mitigates risks to fundamental rights, ensuring AI systems align with EU values.
Risk Management in Critical Digital Infrastructure (pending): Delivers tailored methodologies, use cases, and hazard taxonomies to manage AI risks in critical systems, complementing prEN AI Risk Management.
AI-Enhanced Nudging (JT021003): Addresses ethical risks of AI-driven nudges, safeguarding vulnerable groups and preserving public trust.
Competence Requirements for AI Ethicists (JT021019): Defines core skills and knowledge for AI ethicists to ensure effective ethical integration in AI systems.

For European society, it supports broader adoption of interoperable and transparent digital tools aligned with the Metaverse and Digital Product Passport strategies.It reinforces MPEG’s strategic role in immersive media representation and supports the creation of open and interoperable formats aligned with European values.

Toward the development of EU-friendly solutions for biometrics-based services, employing strong privacy enhancing technologies, thus going further contractual/organisational requirements, to ensure privacy and security by design. Promoting the use of the newest privacy enhancing technologies is in particular very important (biometric technologies are more and more seen as a way to fight against authentication/identification threats in our digital lives) as sharing or leaking biometric information without appropriate protection can be very critical.

The impact is potentially significant worldwide, as the European regulation on AI and its subsequent harmonised standards will be followed by any international company that wants to do business in Europe.