Highest priority risks
Based on confidence and impact
1Unfair capability distributionConfidence 0.71 · High2Undermining creative economiesConfidence 0.71 · High3ToxicityConfidence 0.67 · Medium
View all risks ›AI security records connected to public sources
PromptRiskDB curates and links AI security techniques, risks, mitigations, case studies, vulnerabilities, and sources to help you understand, prioritize, and reduce AI security risk.
4032Total records
170Techniques
1769Risks
44Mitigations
57Case studies
1600Vulnerabilities
326Sources
Browse by category
Featured records
Recently added or updated
AI RisksUnfair capability distribution
View all records ›Unfair capability distribution is an AI risk in 1. Discrimination & Toxicity focused on 1.3 > Unequal performance across groups. It is most relevant during 2...
AI RisksUndermining creative economiesUndermining creative economies is an AI risk in 6. Socioeconomic and Environmental focused on 6.3 > Economic and cultural devaluation of human effort. It is...
AI RisksToxicityToxicity is an AI risk in 1. Discrimination & Toxicity focused on 1.2 > Exposure to toxic content. It is most relevant during 2 - Post-deployment.
AI RisksToxic contentToxic content is an AI risk in 1. Discrimination & Toxicity focused on 1.2 > Exposure to toxic content. It is most relevant during 2 - Post-deployment.
AI RisksStereotypingStereotyping is an AI risk in 1. Discrimination & Toxicity focused on 1.0 > Discrimination & Toxicity; 1.1 > Unfair discrimination and misrepresentation. It...
Case study spotlight
Real-world examples and analyses
Top sources
By record count
- arXiv124
- MIT AI Risk Repository98
- ATLAS (AI Trust & Safety)76
- NIST54
- OWASP38
Built on public sources
arXivMIT AI Risk RepositoryATLASNISTOWASPExplore sources ›Every record is linked to original, publicly available research and documentation.