视觉画布设计
canvas-design
基于设计原则创作静态视觉作品,用于海报与品牌图形。
cn-genai-filing【文档与办公】Use when asked 大模型备案怎么做, 生成式人工智能服务备案还是登记, 算法备案流程, AI 生成内容标识怎么加, 我们接了已备案的大模型还要备案吗, or prepare a generative AI product for launch in mainland China. Produces the obligations that apply (generative AI service filing, registration for apps built on a filed model, algorithm filing, AI-generated content labelling), the documents and security self-assessmen…
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Offering generative AI to the public in mainland China brings obligations under the Interim Measures for the Management of Generative AI Services (effective 15 August 2023), the algorithm recommendation rules, the deep synthesis rules, and the measures for labelling AI-generated content (effective 1 September 2025). Which apply depends on whether the company trains or provides its own model, builds on a model that is already filed, serves the public or only businesses, and what the product generates. This skill works out which apply and prepares the materials.
Write in Simplified Chinese unless asked otherwise. Procedures run through the provincial cyberspace administration (省级网信办) and the national algorithm filing system, and practice changes; confirm with the local 网信办 and counsel. This is preparation, not legal advice.
Ask for these if not provided:
| Obligation | Applies? | Why | Where it is handled | Covering: generative AI service filing (大模型备案) for providers of their own model to the public; registration (登记) for public-facing apps that call a filed model (confirm the local practice); algorithm filing (算法备案) through the national system where the product has public-opinion or social-mobilisation attributes; deep synthesis duties; and AI content labelling (explicit labels users can see, and implicit labels such as metadata).
For filing or registration: service description, model details or the filed model's filing number, the security self-assessment against the national security requirements for generative AI services (TC260 basic requirements), training data and annotation records where relevant, keyword and test question banks, user agreement and privacy policy, and complaint channels.
Where explicit labels appear (text notices, image watermarks, audio cues, video captions), how implicit labels are written into file metadata, and how user downloads and shares keep the labels.
| Item | Owner | Done | Including real-name verification where required, minors protection, a complaint and reporting channel, log retention, and a process for handling illegal content.