Strict rules and governance are the fastest way to slow employees down and kill any desire to experiment – right? You might think so. But current studies prove otherwise: companies with clear AI governance see 18% higher AI adoption and are 13 times more likely to successfully scale their AI initiatives. (IBM) Why? Because broad AI adoption only comes from safe experimentation.

When teams lack a clear framework, one of two things typically happens: either they don't dare touch tools like Claude, Gemini, or Copilot at all, out of fear of making mistakes around data privacy, or they use them quietly in the shadows, leading to uncontrollable "shadow AI." Ethical AI guidelines are the fundamental prerequisite for genuine AI enablement.

As a company that stands for #conscious products, we've gone through exactly this process. Here we share how we developed our internal AI guidelines, built to give our teams the confidence to experiment.

How do we give everyday experimentation the right framework?

There was no single triggering moment for us, no emergency or scandal demanding immediate regulation. The need for ethical and responsible guidelines came directly from the team. As a digital company, we've encouraged our teams from day one to integrate new AI tools into their daily work. Through intensive use, ethical and practical questions kept coming up in everyday situations. What's the actual impact of a complex prompt on our CO₂ footprint? How do we ensure the confidential handling of our own data and our clients' data? The team wanted to take responsibility and needed reliable guardrails to navigate the AI landscape with confidence.

How do you build a compass for technologies that change every day?

A detailed handbook prescribing which button you're allowed to click in Tool X today will be outdated after tomorrow's update. That's why we decided to tie our guidelines to our stable company values. From these, we derived the core of our Ethical AI Guidelines: 5 Guiding Principles that are a reliable north star and point of orientation for our team's work with AI. Our employees can make autonomous, flexible decisions when working with AI, decisions grounded in our company values.

1. Transparency: We're always open about when and how we use AI in our work.
2. Privacy & Safety: We protect sensitive data and our clients' trust.
3. Accountability & Human Oversight: AI assists, humans decide and take responsibility.
4. Fairness & Inclusion: We value diversity and avoid bias in outputs.
5. Sustainability: We use AI purposefully for real value, keep the ecological footprint in mind, and avoid AI "gimmicks" without benefit.

To keep these principles from remaining abstract, we've translated them into "Principles in Action": concrete tips tied to concrete use cases that show our employees how to apply the principles pragmatically in their daily work.

How do you develop guidelines that are embraced by the entire team?

A guideline that expects acceptance can't be written in isolation behind closed doors. At DPM, a colleague for whom "responsibility" is a core concern took ownership of the project and first aligned the broad parameters with management.

But the real breakthrough came through our co-creation approach. We brought engineers, designers and strategists to the table early on to understand real concerns from day-to-day operations. Before the official rollout, we deliberately made time for a workshop session at our strategy day. We discussed the draft openly as a team and adapted critical points together, ensuring the guideline was understood and embraced by everyone.

Why we believe genuine enablement is the best security strategy

The primary goal of our guidelines was to name real risks clearly without discouraging people. By making risks tangible, we reduce fear. On the often sensitive topic of data privacy, for example, we communicated to our employees clearly and simply that certain licensed AI tools can be used under exactly the same permissions and security standards they already know from other tools (e.g., our SharePoint).

This analogy from familiar daily work creates confidence. When the team knows where the boundaries are and which tools are approved, the barrier to trying new things drops, while the company's security level rises significantly.

What we learned (... and would recommend as best practice)

If you're facing the challenge of rolling out AI safely and broadly across your organisation, here are our key learnings from the process:

Want to know more about our eAI guidelines?

Are you facing similar questions in your organisation? Missing the right lever to introduce your teams safely to Generative AI? Let's talk – we're happy to share our learnings, and equally happy to learn from your experiences.