RESPONSIBLE AI CHARTER ยท VERSION [1.0]
Trustworthy AI, as a principle.
Our commitments, applied to every project.
Data protection
Minimization, European hosting, no reuse of your data to train third-party models.
GDPR
EU hosting
Security
Access rights management, encryption, request traceability, and security reviews before production.
Access
Traceability
Transparency
Users know when they're interacting with an AI.
Disclosure
Explainability
Understandable decisions: every result is linked to its sources or driving factors.
Cited sources
Factors
Human oversight
Humans stay in control of sensitive decisions; mandatory validation for risky actions.
Human validation
Fairness
Systematic bias testing before and after deployment.
Bias testing
Frugality
The smallest effective model, to limit cost and footprint.
Frugal AI
AI governance
A use-case registry, a designated owner, and a review committee for every project.
Registry
Committee
AI ACT
How we apply the AI Act
Risk classification for every project, documentation, monitoring. [Legal review required.]