Documented artificial intelligence incidents increased 81 percent from 2024 to 2025, with the AI Incident Database logging 346 incidents during 2025 alone. Cumulative entries in that database reached approximately 800 to 900 unique incidents through the first quarter of 2026, growing at roughly 130 to 180 new incidents per year on a trailing basis.

Analysts tracking the databases caution that the increase reflects two overlapping factors. Deployment of AI systems accelerated across the same period, expanding the population of systems capable of producing an incident. Reporting mechanisms also improved, capturing events that would previously have gone unrecorded. The combination means the figures do not establish a proportional increase in harm per deployed system.

Three public databases track incidents systematically. The AI Incident Database is maintained by the Responsible AI Collaborative. The Organisation for Economic Co-operation and Development runs the AI Incidents and Hazards Monitor, which draws roughly 5,000 to 7,000 incident reports from news monitoring, overlapping substantially with the AI Incident Database while capturing broader media coverage. The MIT AI Risk Repository maintains a third classification. The AIAAIC Repository operates alongside these systems.

The variance in counts across databases reflects differing inclusion criteria rather than disagreement about facts. Media-monitoring systems capture reported allegations, while curated databases apply verification standards before an entry is added.

Source: MIT AI Risk Initiative - https://airisk.mit.edu/blog/ai-incident-tracker-june-2026-update