ModelScan
Community project
ModelScan is an Apache-2.0 open-source scanner from Protect AI for identifying unsafe code in serialized machine-learning model files. It reads supported artifacts without loading them and can run from the command line or in ML and CI/CD pipelines before models are trained, modified, or deployed.
Verified Source: github.com
- Product type
- Community project
- Deployment
- CLIOn-prem
- Organization size
- IndividualSMBMid-marketEnterprise
- Pricing tier
- Free
AI security profile
Open-source scanner that inspects serialized machine-learning model files for unsafe code without loading or executing the model.
Model supply-chain security
Capability checklist
AI Red TeamingHow ModelScan measures up against the full AI Security Testing & Model Assurance taxonomy.
- AI red teaming — not supported
- Runs adversarial tests against models and applications to uncover exploitable or unsafe behavior.
- Prompt-injection testing — not supported
- Tests whether untrusted instructions can override intended system behavior or controls.
- Jailbreak & safety testing — not supported
- Evaluates resistance to policy bypasses, harmful outputs, and other unsafe behaviors.
- Model artifact scanning — supported
- Inspects model files and serialized artifacts for malicious code, tampering, and unsafe components.
- AI supply-chain assurance — supported
- Tracks model provenance, dependencies, integrity, and risk before deployment.
- Continuous AI evaluation — not supported
- Repeats security tests as models, prompts, tools, data, and application behavior change.
Alternatives
Other AI Security Testing & Model Assurance tools with overlapping capabilities, sized for similar teams.