AI Sec Watch Dataset: AI/LLM Security Threat Intelligence

A continuously updated dataset of AI and LLM security intelligence aggregated from 36+ sources including NVD, CISA KEV, GitHub Advisory, arXiv, and cybersecurity news outlets. Each record contains 48 structured fields including LLM-based classification (severity, issue type, attack sophistication, affected AI components) and enrichment from EPSS, CISA KEV, CAPEC, CVSS v3.x, and MITRE ATLAS. The dataset covers October 2012 to present, is updated hourly, and is available in CSV, JSON, and JSONL formats. Platform: https://aisecwatch.com/dataset