Artificial intelligence policy

Our commitments

1. Comply with the EU AI Act, data protection and all other applicable law, and when in doubt, in the strictest way.
2. Transparency: anyone talking to our AI knows it from the start, and what it generates is labelled. These notices cannot be removed.
3. Human oversight: anything sent out on someone's behalf is decided by a person.
4. Safety: if a control fails, the system stops instead of acting without it.
5. Closed intended use: we do not use or offer AI to select or assess staff, score people's creditworthiness, biometrics, emotion recognition or decisions with legal effects on people.
6. Minimal personal data: what goes to cloud models is pseudonymised, and only to providers assessed in writing.
7. Traceability: every use of a model and every automated decision is logged.
8. Training for everyone who operates the AI, ours and our clients'.
9. Honesty: we only claim what we can prove.
10. Continual improvement: every failure is fixed at its root cause and documented.