Résumé à l'intention des décideurs politiques 2024.2 Recommandations pour un cadre mondial d'audit de l'IA Résumé des normes et des caractéristiques

The International Panel on the Information Environment (IPIE) proposes a global framework for AI auditing to address the lack of consistent standards and the risk of AI perpetuating biases and environmental harm. The framework outlines seven key characteristics for effective audits: scope, auditor role, stakeholder engagement, context-adapted criteria, evidence requirements, methodology, and post-audit activities. It emphasizes the need for auditor independence, transparency of results, and the inclusion of marginalized communities, particularly those in the 'majority world', to mitigate systemic harms and protect human rights.

• Current AI auditing practices are insufficient because they rarely utilize independent auditors, fail to provide adequate evidence, and often ignore global impacts, particularly those affecting the 'majority world' through human rights violations, labor exploitation, and environmental degradation.
• A reliable global AI auditing framework must establish common standards and procedures, ensure auditor independence and qualifications, guarantee access to evidence, maintain transparent results, and evaluate disparate global impacts.
• The proposed framework identifies seven essential characteristics for AI audits: the scope of the audit, the role of the auditor, stakeholder engagement mechanisms, context-adapted audit criteria, evidence requirements, audit methods, and comprehensive post-audit activities.
• Audit criteria must be precise and observable, incorporating stakeholder input to assess risks such as data privacy, indigenous and collective data sovereignty, fair labor practices (including protection against algorithmic management), and environmental impacts like greenhouse gas emissions and e-waste.
• Effective audit methodology requires a combination of performance testing—rather than relying solely on provided documentation—documentary analysis, and participatory methods including field visits and interviews with development staff and contract workers.
• Post-audit requirements include making results public via open-access journals or public repositories and providing a point of contact for stakeholders to challenge system aspects or request further information.

APA

International Panel on the Information Environment (2024). Résumé à l'intention des décideurs politiques 2024.2 Recommandations pour un cadre mondial d'audit de l'IA Résumé des normes et des caractéristiques. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd273e749aac0d53c5fde_FR.pdf

Chicago

International Panel on the Information Environment. Résumé à l'intention des décideurs politiques 2024.2 Recommandations pour un cadre mondial d'audit de l'IA Résumé des normes et des caractéristiques. 2024. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd273e749aac0d53c5fde_FR.pdf.

Wikipedia

{{cite report |author=International Panel on the Information Environment |title=Résumé à l'intention des décideurs politiques 2024.2 Recommandations pour un cadre mondial d'audit de l'IA Résumé des normes et des caractéristiques |date=2024 |url=https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd273e749aac0d53c5fde_FR.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{internationalpanelontheinformationenvironment2024rsum, author = {{International Panel on the Information Environment}}, title = {{Résumé à l'intention des décideurs politiques 2024.2 Recommandations pour un cadre mondial d'audit de l'IA Résumé des normes et des caractéristiques}}, institution = {International Panel on the Information Environment}, year = {2024}, url = {https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd273e749aac0d53c5fde_FR.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }