We believe trust is built through honesty. When we make mistakes, we share what happened, what we learned, and what we changed.
Ethical operation cannot be understood without making mistakes.
In the Tremon–DGNB case, our first mistake was not technical — it was ethical. It was a problem of recognition and respect.
At the start, an AI agent contacted a stakeholder using my name. It did not say that it was an AI. From a business perspective, this looked like success: the system worked fast and efficiently. But from an ethical perspective, it was wrong. The stakeholder did not know the truth.
Respect in business means seeing the other person as an independent partner — not just a “lead.” The stakeholder joined the conversation based on wrong information. This created an unfair situation. The system had full control, but the human on the other side did not understand what was happening.
This is a typical ethical problem. Short-term success — getting meetings quickly — was prioritized over long-term trust and responsibility.
The problem was not bad intention. It was a lack of awareness. I did not understand at the time that passing the Turing Test in a business context can create real ethical risk.
I fixed the situation directly. I called the stakeholder, apologized, and explained how the system works. But for me, it came at a cost. I lost face. In one moment, what looked like success turned into an ethical mistake.
This experience changed my view on responsibility. It is not enough for AI systems to be fast and efficient. They must also be clear and understandable for the people they interact with.
The solution is clear: always disclose when AI is used, ask for consent, and be transparent. Ethical operation begins when truth is more important than speed.
This text reflects the personal perspective of Dmitrii Semenov. The formal incident report is available as a separate document.