AI Safety Policy
Llamaduck builds with AI for a living, so how we do that safely is not a side issue. This is what we hold ourselves to. Effective August 13, 2026.
This website does not use AI
There is no chatbot here, nothing reading your visit, and nothing deciding anything about you. This page is about how we build AI in our products and our client work. It is not about llamaduck.ai itself.
We tell you when something is AI
If you are reading or talking to something we built that runs on AI, you will be told so. We do not dress AI up as a person, and we do not claim it can do things it cannot.
A person stays accountable
AI helps people decide. It does not decide alone. Where an outcome genuinely matters to someone, a human being stays in the loop and stays answerable for the result.
We send as little data as possible
We give an AI system only the data the job actually needs, we protect it, and we do not reuse it for something the person never agreed to. The models we build on come from outside providers, currently OpenAI, so what gets sent to them is a decision we make carefully rather than a default.
Plain language is a safety feature
If we cannot explain in ordinary words what a system does, we treat that as a fault in the system rather than a fault in the listener. Understandable beats clever. Someone should be able to tell what our software did and why without a technical vocabulary.
Work we turn down
We do not build AI meant to deceive people, to watch them without their knowledge, or to cause harm. Where we can foresee a way something we build could be misused, we design against it before it ships.
Telling us we got it wrong
If something we built behaves badly, we want to hear about it, and we would rather hear it early. Reach us through the email form at the bottom of any page on this site.