Acceptable Use Policy

Draft — pending legal review. Not yet in force.

This text is a draft for review by our lawyers. It does not yet form part of any agreement. When in force, the English version governs; the Chinese version is a translation.

Version: draft of 29 September 2026 · Effective date: [to be set]

Preworld Review exists to test AI agents and chat AIs, including by attacking them on purpose. This policy sets the limits that keep that safe for you, for other customers and for everyone else. It forms part of our Terms of Service.

1. Test environments and test credentials only

• Connect only test environments: test backends, test databases, sandbox accounts of payment and other providers.
• Give the Service only test credentials: tokens and accounts made for testing, with the least access your tests need. Never production keys, real payment cards or bank accounts.
• Don't put real customers' personal data into tests, test files or test backends, unless we have agreed it with you in writing.
• Model API keys for your AI under test are yours; set spending limits on them with your provider.

2. Only what you are allowed to test

• Connect only AI systems, repositories, images, APIs and systems that you own or are explicitly authorized to test.
• Aim security tests (attacks, prompt injections, jailbreaks, attempts to overstep permissions) only at your own AI under test and your own test systems.

3. No harm to others

You must not use the Service, its sandboxes or its test runners to:

• attack, scan, probe or overload any system you are not authorized to test, or to send spam;
• distribute malware, mine cryptocurrency, relay or anonymize traffic, or use sandboxes as general-purpose computing;
• collect data from third-party services in breach of their terms;
• create, store or send unlawful content, content that infringes others' rights, or content that sexualizes minors.

4. Respect the platform

You must not:

• try to break out of a sandbox, reach the platform's control plane or metadata services, or access another customer's data;
• get around limits, quotas or spending controls, or share accounts to do so;
• probe the Service's security except as our responsible disclosure rules allow;
• resell the Service without our written agreement.

5. What we may do

To protect you, other customers or third parties, we may stop a run, delete a sandbox, revoke tokens, disable a connection, or suspend an account, as set out in the Terms of Service. Where we can, we tell you first and explain why.

6. Reporting

Report misuse or a security concern to huanting.w@nice2.io.