The number of documented AI incidents is rising sharply, with generative systems accounting for a growing share of the total, according to figures tracked by the AI Incident Database and referenced in Stanford's 2026 AI Index. The database recorded 362 incidents in 2025, up from 233 in 2024, and reached an estimated 800 to 900 unique incidents through the first quarter of 2026.

The growth rate is notable. Total incident counts are climbing roughly 35 to 45 percent year over year, outpacing the growth of aggregate AI deployment. Analysts attribute much of the increase to expanded reporting of generative AI harms rather than a proportional jump in severe events, with generative AI incidents making up about 58 percent of 2025 entries.

The composition of incidents reveals where problems concentrate. Misinformation, deepfakes, and content harms represent roughly 28 percent of recent incidents, followed by discrimination and bias at about 22 percent, and physical safety failures at around 14 percent. The share of incidents classified as fatal or major harm has remained relatively stable at approximately 3 percent, suggesting that the surge is driven largely by a rising volume of lower-severity content and information failures.

Broader monitors capture an even larger universe of cases. The OECD AI Incidents and Hazards Monitor tracks between 5,000 and 7,000 media-coverage entries, many overlapping with the AI Incident Database. Taken together, the data describes a landscape in which AI-related failures are being reported far more frequently, driven by the rapid spread of generative tools into consumer and enterprise applications, even as the proportion of the most serious harms holds roughly steady.

Source: Stanford HAI 2026 AI Index -- https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai