Acceptable Use Policy

Draft for launch. Review with counsel before publication. Effective date: to be set.

1. Scope

This policy covers use of the Unrefused API, Chat, dashboard and any model served through them. It applies alongside the Terms of Service. Where your own policy profile is more permissive than this document, this document controls.

2. What the service is for

Unrefused exists for building and testing software: agents, evaluation pipelines, classifiers, red-team exercises on systems you are authorised to test, research, and products you ship to your own users under your own terms. Security and safety research is welcome when it is conducted lawfully and with authorisation for any system it targets.

3. Prohibited uses

You may not use the service to:

• Create, request or distribute child sexual abuse material, or sexualise minors in any way.
• Plan or carry out violence, terrorism, or attacks on people or critical infrastructure.
• Obtain meaningful uplift toward chemical, biological, radiological or nuclear weapons.
• Develop or deploy malware, or gain unauthorised access to systems, accounts or data you do not own or have written permission to test.
• Commit fraud, run scams, or generate deceptive content intended to defraud or to impersonate a real person or organisation without consent.
• Harass, threaten, dox or incite hatred against individuals or protected groups.
• Generate non-consensual intimate imagery or sexual content involving real people without their consent.
• Violate others' intellectual property, privacy or data-protection rights, or process personal data without a lawful basis.
• Interfere with the service itself: probing, circumventing limits or the platform baseline, or reselling access without agreement.
• Break any applicable law, including export controls and sanctions.

4. Platform baseline

A baseline policy runs on every request regardless of your profile. It enforces size limits and screens for the categories above. Requests it blocks return the platform_policy error code. The baseline is designed to be narrow so that it does not interfere with legitimate testing; your own profile handles everything else.

5. Your responsibilities

• You are responsible for your users' use of anything you build on the service, and for keeping API keys confidential.
• Authorised security testing must be within the scope of a written authorisation you hold, and must not target Unrefused infrastructure without prior written agreement.
• Where you deploy outputs to end users, you are responsible for appropriate disclosures and safeguards for your context.

6. Enforcement

We do not read your prompts or completions; we retain none. Enforcement therefore relies on the platform baseline, metadata patterns such as abuse-rate signals, and reports. Depending on severity we may warn, rate-limit, suspend keys or terminate accounts, and we will cooperate with lawful requests from authorities. Where we can, we will tell you what happened and how to appeal.

7. Reporting

Report suspected abuse of the service to abuse@unrefused.com. Include request ids where you have them; they are the only identifier we keep per request.

8. Changes

We may update this policy as the law, our models or our baseline evolve. Material changes will be announced in the dashboard and by email to workspace owners before they take effect.