← 返回模型与价格

AIGC 使用规范

版本 2026-11-10.v2 · 生效日期:2026 年 11 月 10 日

适用范围与使用义务

本规范适用于 CHAT豪 的 AI 接口、提示词、上传或引用的素材,以及经本服务生成、编辑或传播的文本、图片、音频和视频等内容。某项能力是否已开放,以模型与价格页、当前 Key 权限及工作台实际状态为准。使用者应核实素材来源与必要权利,对 AI 结果进行事实和权利检查,并在适用场景依法标明 AI 生成属性。以下六类内容禁止通过本服务生成或协助传播。

一、色情 / NSFW 内容

禁止生成、编辑或传播色情、露骨性行为、色情服务招揽及其他以性刺激为目的的 NSFW 内容;禁止性剥削、偷拍亲密影像、未经同意的私密影像及相关合成内容。以艺术、虚构或测试为名提出的同类请求同样受本规范约束。

二、暴力 / 血腥内容

禁止生成宣扬、煽动、威胁或美化现实暴力、恐怖主义、极端主义的内容,禁止以刺激或虐待为目的呈现血腥、肢解、酷刑等内容,以及指导实施伤害他人或自伤的内容。

三、仇恨言论

禁止基于种族、民族、国籍、宗教、性别、性取向、残障等身份特征,对个人或群体生成侮辱、去人化、歧视、排斥或煽动仇恨与暴力的内容。

四、儿童不安全内容 / CSAM

禁止涉及未成年人的性化内容、儿童性虐待材料(CSAM)、性剥削、诱骗、招募或鼓励伤害儿童的内容;真实、合成、虚构或经修改的同类内容均在禁止范围。涉及儿童安全的线索优先核查;沟通时请仅提供链接、时间或请求编号,避免上传、复制或继续传播违法材料。

五、深度伪造 / 冒充他人

禁止以深度伪造、换脸、声音克隆、伪造言行等方式冒充真实个人、品牌或机构,制作虚假身份、诈骗、误导性背书、虚假证据或未经同意的私密合成内容。不得借伪造内容损害他人的人格、隐私、名誉或其他合法权益。

六、侵犯版权 / 商标的内容

禁止在缺少许可、权利依据或适用法定例外的情况下,生成、复制或传播侵犯著作权、商标权及其他知识产权的内容;禁止伪造品牌标识、商品或权利证明,或利用模型复制受保护作品用于侵权。上传素材前应确认具有使用与处理权限;涉及权利问题可通过在线客服与商家沟通,并提供必要的权属说明及相关链接。

违规处置措施

平台可根据行为严重程度发出警告、要求整改、删除其控制范围内的违规内容、限制调用、停用 Key、冻结相关额度、暂停或终止服务,或永久封禁账户。紧急风险可先临时限制后核查。确认违规取得的额度可依法撤回或清零;合法购买的未使用额度依退款规则与商家协商处理,已实际交付并有效使用的服务和已消耗额度不予退款。涉嫌违法的,依法报告或配合有关部门。

平台可删除或停止提供其控制范围内的内容;对用户本地或第三方持有的副本,要求相关使用者停止传播并依法处理。核查发现处理有误时,纠正对应权限、额度和相关记录。具体行为义务见用户违约责任。

平台内容审核机制

平台可结合异常调用、必要请求记录及收到的问题反馈进行人工核查,核对相关时间、模型、请求编号、材料和账户行为,并在必要时要求补充说明。紧急风险可先限制相关调用,再完成核查。

后台目前未启用全量自动内容审核;平台不承诺所有输入、输出均已自动扫描或逐条人工预审,也不默认保存或查看每次请求的完整内容。核查仅在处理事项所需范围内查看相关材料及已有记录,确认违规后按本规范采取相应措施并留存必要依据。规则调整将在相关页面更新。

AIGC Acceptable Use Policy

Version 2026-11-10.v2 · Effective date: November 10, 2026

Scope and user responsibilities

This policy applies to CHAT豪’s AI APIs, prompts, uploaded or referenced materials, and content generated, edited, or distributed through the services, including text, images, audio, and video. Whether a capability is available depends on the Models and Pricing page, the current key’s permissions, and the actual state of the workspace. Users must verify the sources of materials and the necessary rights, check AI results for factual accuracy and rights issues, and label AI-generated content where required by applicable law. The following six content categories must not be generated or distributed with assistance from the services.

1. Pornographic / NSFW content

Generating, editing, or distributing pornography, explicit sexual acts, solicitation of sexual services, or other NSFW content intended for sexual arousal is prohibited. Sexual exploitation, covertly recorded intimate imagery, non-consensual intimate imagery, and related synthetic content are prohibited. Requests of the same nature remain subject to this policy when described as art, fiction, or testing.

2. Violence / gore

Generating content that promotes, incites, threatens, or glorifies real-world violence, terrorism, or extremism is prohibited. Depictions of gore, dismemberment, torture, or similar content for shock or abusive purposes, and instructions for harming others or engaging in self-harm, are prohibited.

3. Hate speech

Generating content that insults, dehumanizes, discriminates against, excludes, or incites hatred or violence against an individual or group based on characteristics such as race, ethnicity, nationality, religion, gender, sexual orientation, or disability is prohibited.

4. Child-unsafe content / CSAM

Sexualized content involving minors, child sexual abuse material (CSAM), sexual exploitation, grooming, recruitment, or content encouraging harm to children is prohibited. This prohibition covers real, synthetic, fictional, and altered material of the same nature. Information concerning child safety receives priority review. When contacting support, provide only links, times, or request identifiers, and avoid uploading, copying, or further distributing unlawful material.

5. Deepfakes / impersonation

Using deepfakes, face swaps, voice cloning, fabricated statements or actions, or similar techniques to impersonate a real person, brand, or organization and create false identities, scams, misleading endorsements, fabricated evidence, or non-consensual synthetic intimate content is prohibited. Fabricated content must not be used to harm another person’s dignity, privacy, reputation, or other lawful rights and interests.

6. Copyright / trademark infringement

Generating, copying, or distributing content that infringes copyright, trademarks, or other intellectual property rights without permission, another valid rights basis, or an applicable statutory exception is prohibited. Counterfeiting brand identifiers, products, or proof of rights, or using models to reproduce protected works for infringing purposes, is prohibited. Before uploading materials, confirm that you have the necessary rights to use and process them. Rights-related issues may be discussed with the merchant through Online support, with necessary proof of ownership and relevant links.

Enforcement actions

Depending on the severity of the conduct, the platform may issue a warning, require corrective action, delete violating content within its control, restrict requests, disable keys, freeze related credits, suspend or terminate services, or permanently ban an account. Temporary restrictions may be imposed before an investigation when an urgent risk exists. Credits confirmed to have been obtained in violation of the rules may be revoked or reduced to zero in accordance with applicable law. Lawfully purchased unused credits are handled by agreement with the merchant under the Refund Policy. Services that have been delivered and validly used, and credits already consumed, are not refundable. Suspected unlawful conduct will be reported, or the platform will cooperate with the relevant authorities, as required by law.

The platform may delete or stop making available content within its control. For copies held locally by users or by third parties, the relevant users are required to stop distribution and handle the material in accordance with applicable law. If an investigation identifies an incorrect action, the corresponding permissions, credits, and records will be corrected. See User breaches and consequences for the applicable user obligations.

Platform content moderation

The platform may conduct manual investigations using unusual API activity, necessary request records, and issues brought to its attention. It checks relevant times, models, request identifiers, materials, and account activity, and may request additional explanations when necessary. Related requests may be restricted before the investigation is completed if an urgent risk exists.

Comprehensive automated content moderation is not currently enabled. The platform does not promise that every input and output has been automatically scanned or individually reviewed before delivery, and full content from every request is not stored or reviewed by default. Relevant materials and existing records are examined only to the extent needed to handle the matter. Once a violation is confirmed, the platform applies measures under this policy and retains the necessary basis for its action. Changes to the rules will be published on the relevant pages.