Documented AI incidents are climbing faster than AI deployment itself, according to tracking data compiled through early 2026. The AI Incident Database recorded roughly 800 to 900 unique incidents through the first quarter, adding 130 to 180 new entries per year and growing 35 to 45 percent year over year.

Generative AI dominates the recent record. Generative systems accounted for about 58 percent of 2025 entries, reflecting how quickly text and image tools moved into public use. By category, misinformation, deepfakes, and content harms made up roughly 28 percent of recent incidents, followed by discrimination and bias at about 22 percent and physical safety failures at about 14 percent.

The severity mix has stayed relatively stable. The share of incidents classified as fatal or major harm sits at roughly 3 percent, indicating that most documented cases involve reputational, financial, or content-related harm rather than physical injury. Separate tracking efforts recorded a comparable trend, noting an 81 percent increase in documented incidents from 2024 to 2025, with 312 incidents in a single year and 23 classified as critical.

Broader monitoring captures an even larger volume. The OECD AI Incidents and Hazards Monitor tracks 5,000 to 7,000 media-coverage entries, many overlapping with the curated databases. Analysts attribute the rising counts to both accelerating AI adoption and improved reporting mechanisms that surface incidents which might previously have gone unrecorded.

Source: AI Incident Database - https://incidentdatabase.ai/