KI Codex
1. Person judges, machine delivers.
AI prepares, structures, and accelerates. We make the decisions, weigh the options, and take responsibility. No result leaves our office without a human reviewing it and making it their own.
2. We think for ourselves first.
Before we write a prompt, we formulate our own thesis. AI may sharpen and challenge it, but not replace it. Those who outsource thinking forget how to think.
3. We check every statement.
Figures, quotes, and sources from AI systems are considered claims until verified by at least two independent sources. Fabricated facts are not a typo, but a breach of trust.
4. We protect data before we use it.
Customer data, personal data, and confidential documents are only transferred to systems whose data processing practices we understand and which comply with the Swiss Federal Act on Data Protection (FADP) and, where relevant, the GDPR. In cases of doubt, we anonymize the data or refrain from doing so.
5. We make the use of AI transparent.
Customers learn where and how we use AI. We label content that is largely generated by AI where this is relevant for trust and contextualization.
6. We take responsibility for the result.
"The AI wrote it that way" is not an answer. We have reviewed, understood, and can explain anything that appears under our name.
7. We invent nothing and we do not deceive.
No fabricated references, no artificial testimonials, no deepfakes, no imitation of real people without their consent. Our impact comes from substance.
8. We pay attention to distortions.
AI systems reproduce patterns from their training data. We scrutinize results for stereotypes, biased perspectives, and blind spots, especially in HR topics, target group analyses, and evaluations.
9. We retain our judgment.
We deliberately plan tasks that we solve without AI and regularly reflect on where the machine takes work off our hands and where it costs us skills. Efficiency must not erode competence.
10. We remain open to learning and will review this code.
Technology, law, and the state of knowledge change. We review the code at least twice a year, openly document errors, and learn from them.
The machine is artificial. Intelligence must come from humans.