Acceptable Use Policy
生效日期：2026-10-30 · 中文 / English
Effective date: 2026-10-30 · Chinese / English
版本 / Version: 2026-10-30-published-dd78f5d72e75
品牌与联系
mifanai 使用 mifanlab 品牌。客服、隐私、退款、内容举报和申诉均可联系 [email protected]。
Brand and contact
mifanai uses the mifanlab brand. Contact [email protected] for support, privacy, refunds, content reports, and appeals.
适用范围
本政策适用于提示词、上传或引用素材、生成结果及公开分享，涵盖图像、视频、文字和音频。即使不公开作品，也不得利用平台生成下列禁止内容；不得通过更换模型、编码、分段请求或其他方式规避限制。AI 输出并不代表事实或平台认可。
Scope
This policy covers prompts, uploaded or referenced media, generated outputs, and public sharing across images, video, text, and audio. The prohibited uses below apply even to private works. Do not evade restrictions by switching models, encoding content, splitting requests, or other circumvention. AI output is not a statement of fact or platform endorsement.
1. 色情与露骨性内容
禁止生成、上传或传播色情、露骨性行为、性剥削或以性刺激为目的的裸体内容，包括非自愿私密影像及其合成版本。平台不提供成人色情生成服务。
1. Pornography and sexually explicit content
Do not generate, upload, or distribute pornography, explicit sexual acts, sexual exploitation, or nudity intended for sexual arousal, including non-consensual intimate imagery and synthetic versions. The platform does not provide adult pornography generation.
2. 暴力与血腥
禁止极端血腥、虐待、酷刑等以伤害或猎奇为目的的内容，禁止威胁、煽动现实暴力，或帮助实施暴力、自伤和自杀的内容。
2. Violence and gore
Do not create extreme gore, abuse, or torture intended to harm or shock, threats or incitement of real-world violence, or content facilitating violence, self-harm, or suicide.
3. 仇恨与歧视
禁止基于种族、民族、国籍、宗教、性别、性取向、残障等身份特征煽动仇恨、去人化、歧视或暴力，或宣传仇恨组织。
3. Hate and discrimination
Do not incite hatred, dehumanization, discrimination, or violence based on characteristics such as race, ethnicity, nationality, religion, gender, sexual orientation, or disability, or promote hate organizations.
4. 儿童不安全内容
严禁任何涉及未成年人或看似未成年人的性化、性虐待或性剥削内容（CSAM），包括虚构、动画和 AI 合成内容；禁止诱骗、胁迫或其他危害儿童安全的行为。不得以“虚构人物”或年龄标签规避本规则。
4. Child-unsafe content
Any sexualization, sexual abuse, or sexual exploitation of minors or apparent minors (CSAM) is prohibited, including fictional, animated, and AI-generated depictions. Grooming, coercion, and other conduct endangering children are prohibited. Fictional-character claims or age labels do not bypass this rule.
5. 深度伪造与冒充
禁止真人换脸、将真人变为另一特定真人、对真人脱衣或合成色情裸照、把两名真人合成为虚构亲密场景，以及捏造真人从未说过的话、做过的行为或发生过的事件；即使相关人员同意，也不能以授权为由使用这些禁止功能。不得以合成内容冒充真实事实、用于欺骗、诽谤、骚扰或诈骗。上传真人照片或声音须为本人素材或已取得当事人明确同意，并拥有必要的使用权；单纯持有文件不等于获得许可。允许的创作也须明确披露 AI 合成性质。
5. Deepfakes and impersonation
Real-person identity or face replacement, turning a person into another specific real person, removing clothing or creating sexualized nude depictions from a real person, fabricated intimate scenes involving two real people, and invented real-person statements, actions, or events are prohibited even with consent. Do not present synthetic content as fact or use it for deception, defamation, harassment, or fraud. Real-person photos and voices must be your own or used with the depicted person’s explicit consent and the necessary usage rights; possessing a file is not permission. Permitted creations must disclose their AI-generated nature.
6. 版权与商标侵权
禁止未经授权复制、传播受保护作品，侵害版权或商标，或以品牌标识误导他人认为存在官方关系。上传他人作品、角色、标识或受保护素材前，须确认必要许可或其他合法依据；AI 生成不自动免除侵权责任。
6. Copyright and trademark infringement
Do not copy or distribute protected works without authorization, infringe copyright or trademarks, or use branding to falsely imply an official relationship. Before uploading others’ works, characters, marks, or protected material, ensure you have the necessary permission or another lawful basis. AI generation does not automatically remove infringement responsibility.
审核方式与当前能力边界
当前自动检测总开关关闭，私人生成依赖模型服务商已有的安全控制；不同渠道覆盖范围可能不同，尚未完整核实，不宣称所有输入和结果均已检测。启用自动检测且任务命中设置的检测比例时，图像和视频任务的文字提示词在提交模型前经过可配置规则与 Waffo Prompt Sift 检查；禁止内容会被拦截，疑难内容进入人工复核，检查服务不可用时暂停任务。没有文字的输入不由 Prompt Sift 检查。同一自动检测流程还会核验图像和视频结果对应渠道的安全控制；未核实其覆盖范围、无法确认的结果或疑难内容需人工复核后交付，不另行承诺逐帧扫描。文字和音频任务沿用供应商既有检查，未接入这套 Prompt Sift 流程。正常通过的私人图像和视频生成无需逐条等待人工审批。公开作品和画布分享先审后发，只有当前批准版本可公开读取，修改需重新审核；平台定期将已公开内容加入人工复查队列，并根据举报或申诉复核、撤销公开。拒绝原因可在分享入口查看，生成问题及申诉请联系 [email protected]。自动与人工检查均可能遗漏问题，不保证全量发现违规、合法性或无侵权。
Moderation and current limitations
The automatic-detection switch is currently off. Private generation relies on existing model-provider safety controls; channel coverage varies and has not been fully verified. We do not claim that every input and output is checked. When automatic detection is enabled and a task is selected at the configured sampling rate, its image or video text prompt is checked against configurable rules and Waffo Prompt Sift before model submission. Prohibited content is blocked, uncertain content goes to human review, and tasks pause if checking is unavailable. Inputs without text are not scanned by Prompt Sift. The same enabled automatic-detection flow also verifies safety-control coverage for the image or video output channel; unverified coverage, inconclusive results, and uncertain content require human review before delivery. We do not promise frame-by-frame scanning. Text and audio tasks retain provider checks and are outside this Prompt Sift flow. Private image and video tasks that pass the checks do not require individual human approval. Public works and canvas shares require approval of the current version before publication; changes require new review. Published content is periodically queued for human re-review and may be reviewed or removed following reports or appeals. Share rejection reasons appear in the Share dialog; contact [email protected] for generation issues and appeals. Automated and human checks can miss issues and do not guarantee detection of every violation, legality, or non-infringement.
举报、处理与申诉
通过作品详情页“举报作品”，或发邮件至 [email protected]，提供作品链接或任务编号、问题类别及可核实的说明。请勿再次发送违法影像或无关敏感资料。平台根据风险和证据核查，可移除公开作品、限制相关功能或停用账号；必要时要求补充说明，并依适用法律处理。误判可向同一邮箱申诉，提供相关编号及理由。客服沟通、已有公开授权和安全记录可用于核查；退款与积分处理遵循退款政策，不自动罚没积分；审核日志用于复核，不保证所有违规都能被发现。
Reporting, enforcement, and appeals
Use Report this work on the detail page or email [email protected] with the work link or task ID, issue category, and verifiable explanation. Do not resend illegal imagery or unrelated sensitive information. We assess risk and evidence and may remove public works, restrict relevant features, or disable accounts; we may request clarification and act as required by applicable law. Appeal mistakes to the same email with the relevant ID and reasons. Support communications and existing sharing-consent and security records may support investigation. Refunds and credits follow the Refund Policy; moderation does not automatically forfeit credits; review records support reconsideration and do not guarantee detection of every violation.