Statement on the Labeling of AI-Generated Content

This Statement applies to AI-generated or synthesized content provided by Jingdian, including text, images, logo graphics, asset combinations, and page content, and is implemented in accordance with the Interim Measures for the Administration of Generative AI Services of the People's Republic of China, the Administrative Measures for the Labeling of AI-Generated and Synthetic Content, and their supporting mandatory national standards.

This Statement governs both the labeling practices followed by Jingdian as the service provider and the labeling obligations of Merchants and users as the publishing party.

Explicit labels: prompts presented within the generated or synthesized content or on the user interaction interface, in text, icon, audio, or other forms that users can clearly perceive, such as annotations stating “Content generated by AI” or “Generated with AI assistance”.

Implicit labels: information embedded in the data or metadata of a content file that must be identified through technical means, typically including the labeling of the generated or synthesized attribute, the responsible entity, and a content number.

Interaction prompts: when generation capabilities are used, the interface clearly states that the results are generated or assisted by artificial intelligence, that the content is not a factual conclusion, and that it is not equivalent to a finished asset ready for direct external placement.

Explicit labels: for text and image results the platform generates and displays, prompts on the AI-generated attribute are given at the preview and download stages. For historical content or modules where the relevant capability is not enabled, the prompts shown on that module's page apply.

Implicit labels and metadata: for export formats that support metadata preservation, we write the generated or synthesized attribute and provenance information into the file. Some export flows that require unified format conversion (for example, re-encoding an image as PNG) rebuild the file; in such cases, the explicit prompts on the page and the task records retained by the platform serve as the basis for labeling and traceability. Task and provenance information is logged within the platform and can be provided upon request.

We will not remove the labeling information carried by existing generation services in order to evade labeling obligations.

Before publishing externally, placing as an advertisement, printing, or otherwise making AI-generated content available to the public, the user shall complete its own review for accuracy, legality, and brand consistency. AI-generated results may be inaccurate, incomplete, or conflict with third-party rights.

The user shall not delete, tamper with, obscure, or forge AI-generation labels, shall not present AI-generated content as genuine photography, live performance, or pre-existing works, and shall not use it for false advertising, imitating another party's trademark or trade dress, or infringing another party's portrait rights, name rights, or copyright.

When AI-generated content is used for advertising, it shall also meet the requirements of advertising laws and regulations on truthfulness, identifiability, and qualifications; the relevant promotional responsibility is borne by the publishing party.

When publishing through channels such as Mini Programs and apps, the additional declaration and labeling requirements that those channels impose on AI-generated content must be observed.

Jingdian performs text and image generation by connecting to third-party large models and image generation services that have completed the corresponding registration or filing procedures, and routes and switches among them according to feature configuration; the specific providers may change as the service is adjusted.

By using the platform's generation capabilities, you engage us to initiate calls to the above services on your behalf and to process the returned results. Your prompts, brand materials, and generated results will not be used to train third-party models unless you separately grant explicit authorization.

The generation capabilities shall not be used to create content that violates the law, infringes others' lawful rights and interests, constitutes false or misleading information, is obscene or pornographic, is violent or terrorist, deepfakes another person's likeness, or is used to circumvent the platform's review and security mechanisms.

For accounts and content suspected of such uses, Jingdian may issue prompts, restrict generation capabilities, delete content, or suspend the service, and will retain the corresponding records.

If you believe that content generated or displayed by the platform infringes your lawful rights and interests, or has not been labeled as required by this Statement, you may submit proof of rights and specific leads through the contact details published on the pages, and we will verify and handle the matter.

This Statement is adjusted as regulations and product capabilities change; after each adjustment, the effective date is published on the pages.