The number of documented artificial intelligence incidents climbed sharply in 2025, according to figures compiled from the AI Incident Database. The database recorded 362 incidents in 2025, up from 233 in 2024, a rise of more than 55 percent in a single year.

The longer trend points the same direction. Reports of AI-related incidents rose about 50 percent year over year from 2022 to 2024, and in the ten months through October 2025, the count had already surpassed the full 2024 total. Early 2026 data continued the pattern, with more than fifty public AI incidents recorded between January 1 and mid-May.

The tracked incidents span a wide range of harms. Categories include harmful AI-generated content, deepfake impersonation, biased facial recognition tied to wrongful arrests, data exposures, and accidents involving automated systems. Human editors review each entry against publicly available sources such as investigative journalism and academic research before it is added to the database.

The maintainers caution that the totals likely understate the real figure. Because the database depends on public reporting, incidents that draw little media coverage, including those in regions with thinner press attention, are probably undercounted. That means the rising curve reflects both a genuine increase in incidents and improving documentation as more organizations and researchers track them. The data offers one of the clearer quantitative views of how frequently AI systems are linked to public harms, and the trajectory has moved consistently upward as AI deployment has widened.

Source: Our World in Data - https://ourworldindata.org/grapher/annual-reported-ai-incidents-controversies