Acceptable use policy

[Draft: to be approved by legal counsel before publication]

Who this applies to

This policy is part of the agreement between Neural Inverse and each organisation that uses the service we host. It applies to everyone that organisation lets use the service, and to the agents they run in it. The organisation is responsible for its users and agents keeping to it. When you run Neural Inverse yourself, your licence and agreement with us apply instead.

Security and integrity

You may not:

• access, or try to access, another organisation's data, accounts or deployment;
• get around authentication, policy, approvals, rate limits, quotas or the audit log;
• put malware into the service or use it to distribute malware;
• interfere with the service or its other users, including by overloading it on purpose;
• share or sell access, or use credentials that were not issued to you.

Unlawful and harmful use

You may not use the service to:

• break the law, or infringe anyone's intellectual property, privacy or other rights;
• attack, scan or probe systems you are not authorised to test;
• send spam or run phishing or fraud;
• process personal data without a lawful basis, or special category data the agreement does not allow.

AI models and agents

• Follow the terms and usage policies of each model provider you use through the service.
• Keep a person responsible for agent actions: do not configure agents to act on systems where your own policy, law or regulation requires a person to decide.
• Do not use the service to develop weapons capable of mass casualties, or to cause serious harm to people or critical infrastructure.
• Do not use model output to train a model that competes with the provider whose terms forbid it.

Export and sanctions

You may not use or give access to the service in breach of export control or sanctions laws; see export control.

Security testing

You may test the security of your own deployment within the limits of the vulnerability disclosure policy. Testing that could affect other customers, or load testing of the service we host, needs our written agreement first.

What we do about misuse

If we reasonably believe this policy has been broken, we may remove content, suspend the users, agents or credentials involved, or suspend the service, as the agreement allows. Where we can, we tell the organisation's administrator first and give them the chance to fix it; we act straight away when the security of the service or other customers is at risk, or the law requires it.

Reporting misuse

Report misuse to legal@neuralinverse.com, and security vulnerabilities to security@neuralinverse.com.