Responsible AI Collaborative (Partnership on AI)
An open database recording incidents and harms caused or narrowly averted by artificial intelligence systems. Documents each case by description, actors and consequences, from reports and public sources. A reference for real AI loss experience and the construction of risk scenarios.
A collaborative listing based on reports and community curation, with each incident documented and tied to sources. Coverage depends on reporting; it captures the reported and publicised rather than the full universe of incidents.
Use to document the real loss experience of AI systems and build AI E&O or liability risk scenarios. A reference dataset for articles on algorithmic risk.