You own what we deliver
Code, procedures, documentation, and data remain the client’s property.
What we build must be verifiable.
For decision-makers, regulators, and international organizations: our principles of conduct, how we handle data, and the frameworks that guide our work.
Principles
Four verifiable commitments.
Transparency by design
No black box: architecture, responsibilities, and controls are documented.
Governance from day one
Risks, controls, and human review are defined before deployment, not after.
Proof before promise
We publish no result without a source, an owner, and a measurement method.
Data and hosting
Your data stays within the agreed scope.
Chosen hosting
Depending on data sensitivity and the applicable jurisdiction: with you or with a host selected together.
Logged access
Access to our clients’ data is attributed to named individuals, limited, and logged.
No implicit use
A client’s data is never used to train a model without an explicit decision.
Reversibility
Open formats and export procedures: you can change provider.
Frameworks
The frameworks that guide our work.
We use these frameworks as guides. We do not claim any organizational certification that is not proven by an official document.
| Framework | Use at DOSOBE |
|---|---|
| ISO/IEC 42001 | Reference for the governance of the AI systems we design and deploy. |
| Personal data protection | Rules applied to every project that processes personal data, from the design stage. |
| AI regulations | Classification of uses by risk level and associated controls. |
| Digital accessibility (WCAG) | The basis of our audits and our inclusive design. |
Report a problem
A vulnerability, a doubt, a question?
Écrivez à [email protected] avec l’objet « Sécurité ». Nous accusons réception sous deux jours ouvrés.