ml-security - Secure ML systems across the full lifecycle
Secure ML systems and pipelines through threat assessment, robustness testing, privacy protection, artifact security, and compliance controls.
Tags
Updated: 2026-10-08Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Assess ML security threats
- Test adversarial robustness
- Detect data poisoning
- Defend against model extraction
- Defend against model inversion
- Apply differential privacy
- Secure federated learning
- Encrypt and sign models
- Verify ML artifacts
- Secure ML supply chains
- Validate ML API inputs
- Detect and anonymize PII
- Configure audit logging
- Harden model serving
- Audit model fairness
- Map OWASP ML risks
- Map MITRE ATLAS threats
- Manage ML secrets
- Scan dependencies
Inputs
- ML systems
- Models
- Training datasets
- Model artifacts
- ML API endpoints
- Model registries
- Dependencies
- Compliance requirements
- Threat models
Outputs
- Security findings
- Robustness test results
- Privacy-preserving models
- Signed and verified artifacts
- PII detection results
- Anonymized data
- Audit logs
- Compliance mappings
- Hardened serving configurations
Requirements
- Access to ML systems and artifacts
- Access to relevant datasets
- Permissions for model registries and APIs
- Compatible ML framework