Back to Risk Catalogue

OWASP ML Security Top 10

The ten most critical security risks for machine learning systems, published by OWASP.

ML02:2023 Data Poisoning Attack

Risks

AIR-SEC-009Data Poisoning

ML05:2023 Model Theft

Risks

AIR-SEC-008Tampering With the Foundational Model

ML06:2023 ML Supply Chain Attacks

Risks

AIR-SEC-008Tampering With the Foundational Model

ML07:2023 Transfer Learning Attack

Risks

AIR-OP-005Foundation Model Versioning

ML08:2023 Model Skewing

Risks

AIR-OP-014Inadequate System Alignment

ML09:2023 Output Integrity Attack

Risks

AIR-OP-004Hallucination and Inaccurate Outputs

ML10:2023 Model Poisoning

Risks

AIR-SEC-008Tampering With the Foundational Model AIR-SEC-009Data Poisoning