[Google AI Overview at the query, signed out and incognito (operator defaults of 2026-10-01 and 2026-10-14), 2026-10-29. The paste is the Overview expanded (the 'AI Mode Conversation' view, with the source cards that follow the answer); the 'You said' line carries the query twice, run together, and is given once here. Inline links are Google redirect URLs, kept as pasted. The closing disclaimer cut.] [QUERENT] what counts as an incident in the ai incident database [SYSTEM] In the [AI Incident Database (AIID)](https://www.google.com/goto?url=CAESYgHrOzAVoFaATyoT135GKwbgq1JTDhfjO5oDwhNCRt79QowTeOLgpXR1P-Ml7G3U6TsvpiFS6fFlQKDaG1bB24O5tgmNHxARRp3XcX-TPcfQmF7jmShB8mWk3NVW0MDhCtNr), an incident is defined as an alleged harm or near-harm event to people, property, or the environment where an AI system is implicated. [1] To keep the dataset structured and avoid arbitrary tracking, the database uses an adaptive, specific set of criteria outlined in its [AIID Editor's Guide](https://www.google.com/goto?url=CAESYgHrOzAVoFaATyoT135GKwbgq1JTDhfjO5oDwhNCRt79QowTeOLgpXR1P-Ml7G3U6TsvpiFS6fFlQKDaG1bB24O5tgmNHxARRp3XcX-TPcfQmF7jmShB8mWk3NVW0MDhCtNr). For a report to be accepted as a true incident record, it must meet three core baseline thresholds: [[1], [2]] 1. The Core Criteria for Ingestion * Specific Event: It must describe a distinct, real-world occurrence with clear details on what happened, to whom, and where. Theoretical vulnerabilities do not qualify. [1] * AI Implication: The AI system must have played a meaningful role, where the harm likely would not have occurred without the AI's output, omission, or use. [1] * Harm or Near-Harm: The event must result in an adverse effect or a credible chain of events nearly producing one. [1] 2. Common Categories of AI Harms Tracked Using frameworks like CSET's taxonomy, incidents are grouped into categories such as physical safety, discrimination and fairness, financial loss, privacy and security breaches, and misinformation or scams. [[1], [2], [3], [4], [5], [6], [7]] 3. Key Distinctions: "Incidents" vs. "Issues" vs. "Variants" The database differentiates entries based on event status: [1] * AI Incident: A real event where AI caused or nearly caused harm. * AI Issue: A potential harm or vulnerability that has not yet materialized in the real world. * AI Incident Variant: A separate occurrence involving the same AI system, similar causes, and matching outcomes as an existing record. [1] Are you trying to submit a specific case to the database, or are you researching a particular type of AI safety failure (such as LLMs, deepfakes, or robotics) for your own project? MIT AI Risk Initiative