Our Principles
The fundamental principles upon which the ModelTrust platform is built.
Our Core Philosophy
"Trust is not a claim, it's a measurement."
Our platform provides concrete, measurable risk analyses.
Transparency
Our evaluation metrics and analysis results are completely open.
- License information is clearly displayed.
- Detected risks are not hidden.
Measurability
Our risk analysis is based on measurable and verifiable data.
Independence
ModelTrust is completely independent of any model developer.
Explainability (XAI)
We detail the sub-metrics for each score.
Technical Security and Design
Our core design decisions that ensure the security of the platform.
- No-Execution Principle: Model code is never executed during static analysis.
- Isolated Environment: Dynamic tests are performed in completely isolated sandbox environments.
Legal Responsibility
We clearly define our legal boundaries and user responsibilities.
- The terms of service (TOS) of model providers are not violated.
- Disclaimer: The scores we provide are recommendations.