AI Use Policy

1. Scope

This policy explains how the artificial intelligence features of NeuralOS work (agents, chat, automations, generation of code, text, images, video, and voice, and the ATLAS assistant on our website) and the rules for using them. It is part of the Terms of Service and is complemented by the Acceptable Use Policy, the Privacy Policy, and the Security page.

2. Models and providers

NeuralOS does not train its own foundation models: it orchestrates models from specialized providers and picks the most suitable one for each task.

If you connect your own keys (BYOK), calls are made with your provider account and are also governed by your terms with that provider. Details for each provider are on Subprocessors. The list may change as better models become available; we will announce changes as described on that page.

3. The ATLAS assistant on our website

ATLAS, the assistant on the NeuralOS public website, first answers with a local engine that searches our own content without calling any external model. Only when that engine cannot find an answer may ATLAS send your question to a language model through OpenRouter to generate one.

• ATLAS always presents itself as an AI, not as a person.
• Its answers are for guidance only and are not legal, financial, or professional advice; our legal documents prevail.
• Do not share sensitive personal data, passwords, or keys with ATLAS.

4. How we handle your data in AI

• No training on your private data: Klinworks, LLC does not use the private content of your workspace (prompts, files, conversations, outputs) to train or fine-tune AI models.
• Providers: we send providers only what is needed to generate each response. We use their business API services, whose current terms exclude using that data to train their models. Some may retain it for a limited time to detect abuse, under their own policies.
• Agent context: an agent only sees the knowledge sources, files, and integrations you grant it.
• Protection: keys and credentials are encrypted as explained on Security.

Details on what data we process, for how long, and your rights are in the Privacy Policy.

5. Ownership of inputs and outputs

• Inputs: you remain the owner of everything you provide (text, files, code, images, data).
• Outputs: to the extent permitted by law, the outputs you generate with AI belong to you, and we assign them to you without reservation. You may use them for commercial purposes.
• Limits: other people may obtain outputs similar to yours, and in some countries content generated entirely by AI may not be protected by copyright.
• Your responsibility: you must have the right to use what you provide, and you are responsible for how you use outputs, including not infringing third-party rights.

6. Limitations of AI

AI is probabilistic. Its outputs may:

• contain errors, made-up facts, or nonexistent citations ("hallucinations");
• be outdated or incomplete;
• reflect biases present in the model's training data;
• vary from one run to the next for the same instruction.

That is why you should review outputs with human judgment before using them, especially code before deploying it and any content you will publish or send. AI outputs are not a substitute for professional advice (medical, legal, financial, or otherwise).

7. What matters waits for your signature

NeuralOS agents can act for you, but you decide how far:

• Permissions: you choose which tools, data, and integrations each agent can use.
• Approvals: consequential actions (such as spending, sending to third parties, or irreversible changes) can be configured to wait for your explicit approval before they run.
• Traceability: every approval or rejection is recorded with who made it, and relevant actions are written to an audit log that cannot be modified.

You are responsible for the actions you approve or configure to run automatically.

8. Transparency

• When you talk to ATLAS or to a NeuralOS agent, you know it is an AI.
• If you deploy an agent or chat that interacts with other people, you must tell them they are talking to an AI, unless it is obvious from the context.
• Do not present an AI as a real person or hide its use where doing so could deceive or harm someone.

9. Synthetic content and labeling

Images, video, audio, and voices generated with AI are synthetic content. When you publish or distribute it:

• label it clearly if it could be mistaken for real content, especially if it depicts real people, places, or events or deals with matters of public interest;
• do not remove watermarks, provenance metadata, or notices that identify it as AI-generated;
• obtain consent from any real person whose voice or likeness you reproduce.

These rules follow the transparency obligations of the EU AI Act and deepfake rules in other jurisdictions.

10. Prohibited uses

Prohibited uses of AI (fraud, non-consensual deepfakes, content that harms minors, disinformation, malware, circumventing safeguards, among others) are detailed in the Acceptable Use Policy. You must also follow the usage policies of the model providers listed above.

11. High-risk uses and the EU AI Act

If you use AI in sensitive areas (health, legal advice, finance, insurance, employment, housing, credit, or education):

• maintain qualified human oversight of every output that affects a person;
• do not make automated decisions with legal or similarly significant effects without human review;
• tell affected people that AI was used.

NeuralOS is a general-purpose tool. If you use it to build a system that the EU AI Act considers high-risk, you are the one who must meet the obligations that apply to you as a provider or deployer. Practices prohibited by that Act are excluded in the Acceptable Use Policy.

12. Reporting a problem

If the AI produces an output that is harmful, offensive, seriously inaccurate, or infringes rights, or if you notice misuse, email ceo@neuralos.io with the context, the approximate date, and, if you can, a screenshot. We review every report and use it to improve our safeguards.

13. Changes to this policy

AI moves fast, and so will this policy. The last-updated date appears at the top. If a change is material, we will give you at least 30 days' notice, unless we must act sooner for legal or safety reasons.

14. Contact

For any questions about this policy, email ceo@neuralos.io.

Klinworks, LLC · a Delaware limited liability company (USA) · file number 10704723.