Recomendações para uma estrutura global de auditoria de IA

The International Panel on the Information Environment (IPIE) presents a global AI auditing framework designed to address the lack of common standards and the failure of current audits to account for global, disparate impacts. The framework outlines seven key features: audit scope, auditor roles, stakeholder engagement, context-specific criteria, evidence requirements, auditing methodologies, and post-audit activities. It emphasizes the need for auditor independence, transparent results, and the inclusion of environmental and labor impact assessments to protect marginalized communities and the collective well-being.

• Current AI auditing practices are insufficient because they rarely involve independent auditors, fail to provide adequate evidence, and often ignore the disproportionate global impacts of AI, particularly regarding human rights violations, labor exploitation, and environmental degradation.
• An effective and reliable global AI auditing framework must be built upon five core pillars: common standards and procedures, auditor independence and qualifications, access to comprehensive evidence, transparent results, and the assessment of disparate global impacts.
• The proposed framework identifies seven essential features for AI audits: defining the audit scope, clarifying the auditor's role, establishing stakeholder engagement mechanisms, setting context-appropriate criteria, determining evidence requirements, applying auditing methodologies, and conducting post-audit activities.
• Audit criteria must be specific and contextual, covering diverse areas such as indigenous and collective data sovereignty, fair labor practices (including protection against algorithmic management), and environmental impacts such as greenhouse gas emissions, water consumption, and e-waste.
• To ensure validity, auditors should not rely solely on documentation provided by the audited party but must perform active performance and scenario testing over extended periods to capture long-term systemic impacts.
• Post-audit transparency is critical; final reports should be publicly available through open-access media or searchable repositories and include actionable items and a point of contact for stakeholders to contest AI system aspects.

APA

International Panel on the Information Environment (2024). Recomendações para uma estrutura global de auditoria de IA. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd28fddb6c2022d343d4c_PTB.pdf

Chicago

International Panel on the Information Environment. Recomendações para uma estrutura global de auditoria de IA. 2024. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd28fddb6c2022d343d4c_PTB.pdf.

Wikipedia

{{cite report |author=International Panel on the Information Environment |title=Recomendações para uma estrutura global de auditoria de IA |date=2024 |url=https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd28fddb6c2022d343d4c_PTB.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{internationalpanelontheinformationenvironment2024recomendaes, author = {{International Panel on the Information Environment}}, title = {{Recomendações para uma estrutura global de auditoria de IA}}, institution = {International Panel on the Information Environment}, year = {2024}, url = {https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/67abd28fddb6c2022d343d4c_PTB.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }