—
—
Checked daily
—
—
—
—
—
—
—
—
Year by year
—
Latest incidents added
—
| Happened | Incident | Developer named |
|---|
Whose AI
—
| Developer | Incidents, 12 months |
|---|
AI and digital
- AI models and chips →How fast AI is growing, and who builds it.
- Data breaches →Breaches made public.
More
- AI and digital →The topic page, with the latest numbers.
How we know this
- Source: the AI Incident Database, run by the Responsible AI Collaborative, which records cases where AI systems caused or nearly caused harm, as reported in the news and elsewhere; with risk classifications from the MIT AI Risk Repository.
- Update cadence: the database publishes a full snapshot every week; we check every night.
- Methods: we count incidents by the date they happened and by the kind of harm, and count the developers named in the last 12 months. We use titles, dates and tags only.
- Limitations: incidents are allegations, recorded when someone reports them. More incidents can mean more AI in use, more attention or more reporting, as well as more harm. Recent incidents are added with a lag and many are not yet classified.
- Primary data: AI Incident Database · MIT AI Risk Repository · How this page is built