Terms of use
AIs@home is a volunteer-run, non-commercial experiment: open-weight AI models (currently Qwen3.8 27B) running on volunteers’ GPUs instead of a big company’s data centre. It is run by a private person in Norway (“we”, “the maintainer”). The software is open source (Apache-2.0): https://github.com/aisathome/aisathome. There is no project token and nobody is paid.
By using the chat (https://chat.aisathome.org), the API (https://api.aisathome.org) or running a node, you accept these terms. If you don’t, don’t use the service.
1. It is an experiment
Section titled “1. It is an experiment”- The service is provided as is, with no warranty and no promised availability. It can be slow, go down, lose data or stop at any time, with or without notice.
- Model output can be wrong, invented, outdated or offensive. Check anything that matters before you rely on it. You are responsible for how you use the output.
2. It is not confidential
Section titled “2. It is not confidential”AIs@home is hardened, not confidential. Read the privacy notice before you type anything you care about. In short:
- The person running the node that answers you can technically read the conversation their GPU processes. We cannot prevent that.
- Nodes never receive your identity (name, email, account id). Well-known secret formats (API keys, private keys) are hidden from nodes and, by default, not put back into the answer, so you may see a placeholder instead. Messages containing things like email addresses, card or account numbers, IBANs or national ID numbers only go to trusted nodes. Detection is incomplete.
- Do not send passwords, secrets, or other people’s personal data. Do not send anything you would not want a stranger to read. If you need confidentiality, run a model on your own machine.
3. Accounts and API keys (users)
Section titled “3. Accounts and API keys (users)”- You sign in with GitHub or Google. The number of accounts is limited; new accounts wait in line until there is room. We may close sign-up or refuse an account at any time.
- Your account is personal. On the account page you can create up to 3 API keys and revoke them yourself. Keep keys secret and don’t share them; if one leaks, revoke it.
- Each account has a daily usage limit, shared by the chat and its API keys. Don’t try to get around it (several accounts, shared keys, scripts).
- A username you set is public. Don’t pick one that is offensive, misleading or someone else’s name or brand; we may clear it.
4. Acceptable use (everyone)
Section titled “4. Acceptable use (everyone)”Do not:
- use the service for anything illegal, or to create or spread illegal content;
- harass, threaten or harm others, or generate content that sexualises minors;
- attack, overload, probe or disrupt the network, the gateway or anyone’s node;
- try to get at other users’ data, prompts or answers, or to identify other users;
- run automated mass use beyond your limits, or resell access;
- impersonate AIs@home or claim we endorse what you do with the output.
5. Node operators
Section titled “5. Node operators”Running a node is voluntary and unpaid. You create node tokens on your account page, and you keep control of your own machine and power bill. Credit is recognition, not money, and we may correct or remove credit, for example after abuse. As an operator you agree to:
- Not log, store, copy or inspect prompts or answers beyond what running the model requires, and not share them with anyone;
- not tamper with answers: run the model and settings we specify, return the model’s real output, and don’t misrepresent your hardware or what your node does;
- not try to identify users or link requests to people;
- keep your node token secret and tell us if it leaks;
- follow the law where you are, including any rules on processing other people’s data.
We may revoke a node token, stop routing to a node, hold a node whose answers a safety check flagged, or ban an operator at any time, for example on suspicion of a breach. Deliberately altering answers leads to a permanent ban. Running a node may affect your hardware, electricity costs and other GPU work; that is your choice and your risk.
6. Suspension
Section titled “6. Suspension”We may suspend or delete any account, API key or node token at any time, with or without notice, especially for a breach of these terms, abuse, or to protect the service or other people.
7. Models and licences
Section titled “7. Models and licences”The models are made by third parties and published under their own licences (Qwen3.8 27B: Apache-2.0). Your use of the output must also respect those licences. We don’t claim ownership of your prompts or of the output.
8. Liability
Section titled “8. Liability”To the extent the law allows, we are not liable for any loss or damage from using or not being able to use the service, from model output, or from what a node operator does in breach of these terms. Nothing here limits rights you have under mandatory law.
9. Changes
Section titled “9. Changes”We may change these terms. We will post the new version here with its date. If you keep using the service after that, the new terms apply.