Acceptable use policy

1. Scope and conditions for enforcement

This policy describes the boundaries for using the related services. For browsing the website, the parts relating to website security and use of content apply. For rules covering APIs, models, research demos or interactive products, the scope, enforcement process and appeal channels must be specified when the corresponding service is opened.

This policy does not mean that the website has made available every capability it mentions. The specific features available, access conditions and limits of authorization are as set out in the actual product interface and the formal agreements.

2. No assistance with serious harm or unlawful activity

The related services must not be used to plan, carry out or facilitate violence, exploitation, abuse or other serious harm against others. They must not be used to produce or distribute child sexual exploitation material or terrorist propaganda, or to provide actionable support intended to cause real-world harm.

They must not be used for fraud, theft, money laundering or other unlawful activity. Related topics in journalism, research, education and the arts should be assessed in context, so that discussing a topic is not confused with helping to cause harm.

3. Protecting privacy and authenticity of identity

Do not collect, infer, expose or disseminate sensitive personal information without a lawful basis, and do not engage in stalking, harassment or targeted intimidation. When processing personal data, you must have the necessary authorization or another appropriate basis.

Where a real person’s voice, face or other identifying characteristics are involved, the relevant rights must be confirmed. Generated content must not be used to forge authorization, commit fraud, damage anyone’s reputation or impersonate genuine communications. Synthetic content that could be mistaken for real must carry a disclosure appropriate to the context of use and applicable requirements.

4. Minors and vulnerable people

Do not create exploitative content targeting minors, encourage them to engage in dangerous behavior or manipulate them by taking advantage of their age and inexperience. Any interactive experience directed at minors requires prior assessment of content, privacy, relationship boundaries and oversight mechanisms.

Do not exploit users’ psychological vulnerability, financial hardship or dependence on the system to deceive or pressure them. Companion or anthropomorphic applications must respect users’ wish to stop interacting and avoid setting themselves against users’ real-world support systems.

5. Significant decisions and professional contexts

In contexts involving healthcare, law, finance, employment, educational opportunities or anything else that may significantly affect individuals’ rights and interests, the limitations and biases of models and the consequences of errors must be assessed, and oversight by qualified personnel must be arranged as actually required.

Unverified generated output must not be used as the sole basis for a significant decision, and a model’s natural-language expression must not be used to create the impression that it holds professional qualifications or authoritative judgment. What is permitted in any particular context, and under what conditions, must be specified separately when the service is opened.

6. Cybersecurity and service integrity

Do not access systems without authorization, obtain credentials, deploy malicious code, circumvent usage limits or interfere with the services. Do not disrupt availability through automated traffic, repeated requests or other means.

Good-faith security testing must be conducted within the targets and methods expressly authorized. If you discover a vulnerability, minimize the impact on real users and data as far as possible and report it through the security reporting channel officially published by the service.

7. Content rights and transparent use

Users must confirm that they hold the copyright, likeness, trademark and other rights required for the material they input and the content they publish. Purchasing a generation service or obtaining access to a model does not automatically secure licenses for third-party material.

Do not fabricate research evidence, evaluation results, partnerships or brand endorsements. When citing model output or evaluations, keep the necessary version, date and context information, and do not present illustrative content as verifiable fact.

8. Enforcement, appeals and policy revisions

A specific service may, to the extent permitted by its formal agreements and policies, investigate suspected violations, restrict features, suspend access or take other necessary measures. Enforcement should take into account the severity of the conduct, the evidence and the impact on other users, and should not be broader than necessary.

At formal launch, the means of reporting, giving notice and appealing must be published, together with how wrongful decisions are handled and reviewed. The website currently has no means of submitting reports directly, and this page should not be regarded as having established such a mechanism.

This policy will be adjusted as product features, actual risks and applicable requirements change. Significant changes must clearly indicate the version and effective date and be made available for review in a reasonable way.