The AI Incident Database recorded a sharp rise in documented AI incidents in 2025, reflecting both the spread of the technology and growing scrutiny of its failures. The data track real-world cases of AI systems causing harm or controversy.

The database logged 362 incidents in 2025, up from 233 in 2024, a roughly 55 percent increase year over year. Generative AI now accounts for the majority of newly logged incidents, about 58 percent of 2025 entries, reversing the earlier dominance of recommendation-system and computer-vision cases. Within the year's records, a large share involved fake audio, video, or images, and one widely used chatbot appeared in dozens of separate cases.

The rising count reflects more AI deployment and better reporting rather than a single cause, but it signals that incidents are scaling alongside adoption. For organizations deploying AI, the data underscore the value of monitoring, human review, and clear accountability for AI-driven outputs. The database is maintained as a public catalog so that researchers, companies, and regulators can study patterns across documented failures. The 2025 total set a new annual record for logged AI incidents.

Source: AI Incident Database - https://incidentdatabase.ai/