UTC

209. AI security incidents and exercises

Attacks in the wild (ATLAS)27 casesExercises and research (ATLAS)50 casesSecurity vulnerabilities and attacks12 casesPrivacy and data leaks22 casesCyberattacks and mass harm10 casesFraud, scams and manipulation51 casesOther reported incidents37 cases2750225137209AI security incidents and exercises
Watchlist
ATLAS case studies
77
Attacks in the wild
27
Reported in the news
132
In 2026
56
Pinch or Ctrl/⌘ + scroll to zoom, drag to move
Incidents · 27
Exercises and research · 50
Reported in the news · AI Incident Database · 132

Titles from the AI Incident Database (incidentdatabase.ai), snapshot of 2026-11-16, under CC BY-SA 4.0. Incidents that involve generative AI and a security harm only.

From MITRE ATLAS 2026.10 (Apache-2.0). ShieldMap lists the facts and the technique chain of each case study and links to ATLAS for the full write-up; it does not copy the steps.