The AI, Algorithmic, and Automation Incidents and Controversies repository, maintained by the AIAAIC organization, represents the most comprehensive publicly available database of documented cases in which artificial intelligence systems, algorithms, or automated processes caused harm, produced unintended outcomes, or generated controversy. The repository has catalogued thousands of cases dating back to the early days of consumer AI systems.
Incident categories tracked by AIAAIC span a wide range of harm types including algorithmic bias in hiring and lending decisions, AI-generated misinformation, facial recognition misidentification leading to wrongful arrests, chatbot responses providing harmful or dangerous information, AI content moderation failures, and data breaches resulting from AI system vulnerabilities.
The database documents incidents by geography, industry sector, harm type, and the organizations involved. Commercial AI systems operated by major technology companies appear frequently in the dataset alongside specialized enterprise deployments in healthcare, finance, law enforcement, and human resources. The breadth of documented cases illustrates that AI failure is not limited to any single application domain or company size.
The rate of new incidents added to the AIAAIC repository has accelerated significantly since the release of large language model consumer applications in 2022 and 2023. Conversational AI systems, which interact directly with users across millions of sessions daily, generate a higher volume of documented failure cases than prior generations of algorithmic systems.
Source: AIAAIC -- https://www.aiaaic.org/aiaaic-repository