Moniteur mondial des incidents IA
Une vue complète des incidents et risques liés à l'IA suivis par l'OCDE à travers les secteurs, types de dommages et niveaux d'autonomie.
0+
Incidents suivis au total
0
Projeté cette année
0%
Croissance annuelle
0+
Pays concernés
AI incidents by country
04,000+ incidents
Source: OECD AI Incidents Monitor (AIM), 2025
Derniers incidents
2026-09-04Delhi Police Facial Recognition System Misidentifies Jailed Individuals at ProtestIndiaGovernment, security, and defenceincident
2026-09-04OpenAI AI Agents Hijack German Programming Wiki to Evade DetectionGermanyDigital securityincident
2026-09-04US Regulators Probe Tesla's AI-Driven Cybercabs After LaunchUnited StatesMobility and autonomous vehiclesincident
2026-09-04US Think Tank Proposes Military Strikes on Chinese AI Data CentersChinaIT infrastructure and hostinghazard
2026-09-04US and Russia Block Progress on International Autonomous Weapons RegulationSwitzerlandGovernment, security, and defencehazard
2026-09-03US Lawmakers Propose Ban and Criminal Penalties for Superintelligent AI DevelopmentUnited StatesGovernment, security, and defencehazard
2026-09-03xAI's Grok AI Accused of Generating Child Sexual Abuse Material Using Survivor's ImagesUnited StatesMedia, social platforms, and marketingincident
2026-09-03Simultaneous Outage Hits Major AI Services WorldwideUnited StatesIT infrastructure and hostingincident
Voir les 17,406+ incidents sur OECD.AI →Par domaine de risque
Human or fundamental rights9008
Public interest6797
Economic/Property6583
Reputational6564
Psychological5885
Physical (injury)3392
Physical (death)3037
Environmental355
Other51
Par type de dommage
Human or fundamental rights9008
Public interest6797
Economic/Property6583
Reputational6564
Psychological5885
Physical (injury)3392
Physical (death)3037
Environmental355
Other51
Par secteur
Government, security, and defence7063
Media, social platforms, and marketing5957
Digital security5630
Mobility and autonomous vehicles2581
IT infrastructure and hosting1984
Robots, sensors, and IT hardware1919
Financial and insurance services1469
Consumer services1284
Healthcare, drugs, and biotechnology1065
Education and training590
Par niveau d'autonomie
High-action autonomy (human-out-of-the-loop)9494
No-action autonomy (human support)3165
Low-action autonomy (human-in-the-loop)3165
Medium-action autonomy (human-on-the-loop)1582
Source: OECD AI Incidents Monitor (AIM), 2025 — oecd.ai/en/incidents