The number of documented AI-related incidents climbed to 362 in the latest edition of the Stanford Institute for Human-Centered AI's annual AI Index Report, up from 233 the year before and fewer than 100 annually before 2022, according to the report's Responsible AI chapter. The Index tracks publicly reported cases in which AI systems caused harm, malfunctioned, or generated significant controversy.

The report also cites a parallel tracking effort by the OECD's AI Incidents and Hazards Monitor, which recorded a peak of 435 monthly incidents in January 2026 and a six-month moving average of 326, pointing to a sustained rise in reported problems rather than a single spike.

Alongside the incident count, the report highlights a widening gap between the frequency of AI-related problems and organizations' confidence in handling them. A joint survey by the AI Index and McKinsey found the share of organizations rating their AI incident response as "excellent" fell from 28 percent in 2024 to 18 percent in 2025, while the share reporting three to five incidents in the past year rose from 30 percent to 50 percent. The report's authors also note that while most frontier AI developers publish results on capability benchmarks, disclosure of responsible AI benchmarks remains inconsistent across the industry, which limits outside researchers' ability to compare safety performance between systems.

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