About PromptRiskDB
Dataset NotesPromptRiskDB explains AI security records by connecting public datasets and linking back to the original sources.
How it works
How PromptRiskDB connects public AI security records and links back to the original sources.
About PromptRiskDB
PromptRiskDB is a map of public information about AI security. It brings together attack techniques, risks, safeguards, case studies, and software vulnerabilities so readers can follow how one record relates to another.
Each page starts from published datasets. When a technique points to a mitigation, a case study mentions a CVE, or an AI risk belongs to a larger category, PromptRiskDB uses those recorded connections to show the relationship in one place.
How pages are built
The site can make the information easier to browse by adding headings, summaries, and links between related records. It does not invent incidents, risk descriptions, vulnerability details, or source claims. If the underlying public records do not say something, the page should not present it as a fact.
Sources and attribution
Original source
CISA Known Exploited Vulnerabilities
CISA KEV marks CVEs that are known to have been exploited and adds vendor, product, remediation, and ransomware context.
Original source
MIT AI Risk Repository
The MIT AI Risk Repository provides AI risk records, domain taxonomy, causal taxonomy, and publication metadata.
Original source
MITRE ATLAS Data
ATLAS provides the AI security tactics, techniques, mitigations, and case studies used as the central map for these pages.
Original source
MITRE ATT&CK Enterprise STIX
MITRE ATT&CK Enterprise provides the broader cybersecurity technique, tactic, mitigation, and STIX context linked from ATLAS records.
Original source
NVD CVE API 2.0
NVD enriches selected CVEs with severity, status, references, weakness, and vulnerability detail fields.
