Media-reported AI incidents involving content generation reached nearly 500 per month by January 2026, a tenfold increase over six years, according to OECD AI Incidents and Hazard Monitor data charted by Statista.

The growth curve has steepened in stages. Monthly reported incidents in the OECD database averaged about 50 in early 2020, when generative tools were confined to research labs and early adopters. By early 2024, following the mass adoption of text and image generators, the monthly average had climbed past 200. The most recent reading of roughly 500 per month reflects both the explosion of AI-generated content across the internet and improved tracking of the harm it can cause.

Content generation incidents span a wide range: fabricated news articles and deepfaked public figures, AI-written legal filings containing invented citations, chatbots dispensing dangerous advice, and brand advertising that misfires in public. The OECD monitor aggregates media reports rather than confirmed regulatory findings, so the figures track public visibility of AI failures as much as the underlying event rate.

The direction is what matters for companies deploying the technology. Incident volume is compounding at the same pace as adoption, and every category of content failure documented in the database traces back to systems that shipped without adequate review of what they could produce.

Source: Statista - https://www.statista.com/chart/35846/ai-incidents-involving-content-generation/