Advancing Trustworthy AI for Sustainable Development: Recommendations for Standardising AI Incident Reporting

Standardized reporting is urgently needed: The paper identifies that current AI incident databases lack consistent protocols for collecting incident data, creating gaps that prevent organizations from learning from past problems and developing effective prevention strategies.

Nine specific gaps were identified in existing systems: The research systematically analyzed open-access AI-incident databases and found nine distinct problems in how incidents are currently reported, suggesting that ad-hoc approaches are insufficient for managing AI risks at scale.

Better incident data enables prevention: By establishing standardized reporting practices, organizations and society can more reliably identify patterns in AI failures, allowing stakeholders to develop strategies that prevent similar harms from occurring in the future.

Trustworthy AI supports broader development goals: The paper connects responsible AI practices to the UN Sustainable Development Goals, arguing that standardized incident reporting is essential infrastructure for ensuring AI technologies contribute to sustainable and inclusive digital transformation rather than causing harm.

International cooperation is essential: The authors emphasize that addressing AI incident reporting gaps requires coordination among multiple stakeholders across countries, suggesting that no single organization can solve this problem alone and that global standards are necessary.