AI Failure Index

The AI Failure Index is a systematic catalog or measurement framework that tracks instances where AI systems produce incorrect, harmful, or unintended outputs in real-world deployments. Organizations use these indices to document patterns of AI failures across industries, understand root causes, and identify which model types or application domains carry the highest operational risk. For enterprise governance teams, maintaining visibility into failure rates and types is essential for establishing accountability mechanisms, setting performance thresholds, and demonstrating due diligence to regulators and stakeholders.

1 item

ResearchGlobal2026-08-06

Hallucinated Threat Report Blocked a Startup's Domains Worldwide

The AI Failure Index documented a case in which a security threat report, allegedly generated with the assistance of a large language model, branded a startup as a Chinese spy front without factual basis. The report was published without expert review or source verification, causing the startup's domains to be blocked globally. The incident illustrates that AI-assisted security intelligence products carry direct third-party harm liability when human fact-checking controls are absent.

hallucination security intelligence defamation human oversight third-party harm threat reporting AI Failure Index