Legal
Copyright and Content Responsibility
Last updated: 9 September 2026
What this means
Geocentric does not host user-published content. There is no upload, no public sharing, and no library of user material — conversations live in your own browser. So there is no designated copyright agent and no takedown queue, and we are not going to publish the machinery of one we do not operate.
What this page does set out is who is responsible for what, and how to reach us if you believe a Geocentric model has reproduced your work.
User responsibility
These points apply to everything you submit to a Geocentric model and everything you do with what it returns:
- You are responsible for your prompts, inputs, uploads, your use of outputs, and the actions you take using the model.
- You must hold the rights and permissions necessary for the content you submit.
- You must comply with applicable law.
- You may not use the service to infringe third-party rights.
- Geocentric does not endorse and does not assume responsibility for users' independent conduct.
- We may suspend or restrict access for violations of these terms.
- Our limitations of liability remain subject to applicable law and do not exclude what cannot lawfully be excluded.
They also appear in the Terms of Service, which is the agreement they form part of.
We do not host user content
The Arc chat service has no upload, no publishing, and no sharing feature. Conversation history is stored in your browser. We retain prompts and responses only where a user opts in, and delete them within 30 days, as described in the Privacy Policy. That retention is internal and is not a public repository of user material.
Because we do not host user-published material, there is nothing for a takedown notice to remove, and we have not designated a copyright agent. If we later add a feature that hosts user content, we will publish the corresponding process before that feature launches.
Model output and copyright
Two things that are frequently assumed and are not true. Text generated by an AI system is not automatically protected by copyright — material lacking sufficient human authorship may not be protectable at all. And generated text is not guaranteed to be free of others' rights: a model can produce text resembling material it was trained on. We do not warrant either way. See the Terms of Service and our AI and Model Disclosure.
Raising a copyright concern
If you believe a Geocentric model reproduces your work, or that material we publish on this website infringes your copyright, write to contact@geocentricai.com. Please include what the work is, where you saw the problem — for a model output, the prompt and the response are the most useful thing you can send — and how to reach you.
We will investigate and respond. Where a concern is substantiated we will act on it, which may include removing material we publish or adjusting how a model is served or trained.
Our training data sources are published on the Training Data Transparency page, so you can see what went into a model before you write to us.