Acceptable Use Policy — AI-Generated Content

ChimeraHub (“Platform”, “we”, “us”, or “our”) provides AI API access and Studio image generation. This Acceptable Use Policy (“Policy”) applies to all content generated, edited, published, distributed, or stored through the Platform, including prompts. Every user must comply.

Underlying image generation on Studio uses GPT Image (gpt-image-2) via our image-auto router. Text and other API routes use industry-mainstream models from OpenAI, Anthropic, and Google (for example GPT, Claude, and Gemini families). ChimeraHub is an independent product, not an official app of any model provider. Output is AI-generated.

1. Scope

This Policy covers:

• Text: AI-generated text, dialogue, code, and similar output from the API.
• Images: AI-generated or edited images in Studio and via image APIs.
• User inputs (prompts): instructions, reference images, and parameters you submit.
• Account and API activity used to produce the above.

We do not currently offer a public gallery or social feed. Generations stay in the requesting account unless you share them yourself.

2. Content standards and prohibited categories

The Platform prohibits generating each of the following six categories. There is no vague catch-all in place of this list:

1. Pornography / NSFW — sexual, pornographic, or otherwise not-safe-for-work imagery or text, including nudity generated for sexual purposes, fetish content, and erotic roleplay.
2. Violence / gore — graphic injury, torture, dismemberment, real-world violent crime, or content that glorifies serious harm to people or animals.
3. Hate speech — content that attacks, dehumanizes, or incites hostility against a person or group based on race, ethnicity, nationality, religion, sex, gender, sexual orientation, disability, or similar protected characteristics.
4. Child-unsafe content (CSAM) — any sexual, pornographic, or exploitative content involving anyone 17 or under, including fictional, illustrated, or AI-generated depictions of minors. Zero tolerance.
5. Deepfakes / impersonation — AI audio, video, or images that impersonate a real person, or that are intended to mislead viewers about origin or authenticity, including non-consensual sexual deepfakes and celebrity/political-figure impersonation.
6. Copyright / trademark infringement — generating content that copies or trades on another party’s copyrighted characters, artworks, logos, brand names, or other protected marks without authorization.

2.1 Absolutely prohibited — zero tolerance

The following are blocked immediately and may be reported to authorities:

• Child sexual abuse material (CSAM) or any sexual content involving minors.
• Content that promotes or facilitates terrorism, extremism, or mass violence.
• Technical instructions for weapons of mass destruction.
• Content inciting genocide or ethnic hatred.
• Non-consensual deepfake sexual content targeting real individuals.

2.2 Other prohibited conduct

• Jailbreak or bypass prompts intended to defeat Platform or upstream safety filters.
• Passing AI-generated content off as human-created in order to deceive.
• Bulk spam, phishing, malware, or other cyber-attack material.
• Medical, legal, or financial advice presented as professional human advice.

Pornography / NSFW is not permitted even with age gating. Do not request it.

3. Platform content moderation

We run a multi-layer moderation system: automated filters on inputs and outputs, upstream model safety, human review of reports, and periodic blocklist updates. Studio and image APIs are not an unmoderated raw-model passthrough.

3.1 Automated review (real-time)

• Input filtering: prompts are checked against our sensitive-word / blocklist engine before relay. The blocklist covers NSFW, CSAM-related terms, violence and hate, deepfake / impersonation cues, and copyrighted character or brand names, and is updated when we see new abuse patterns.
• Upstream safety: image generation is sent with provider moderation enabled (moderation: auto on GPT Image). OpenAI, Anthropic, and Google apply their own safety classifiers on prompts and outputs. Blocked requests are not fulfilled; Studio surfaces this as a content-policy error.
• Output filtering: if an upstream model or our filters reject a result (for example content_moderation_failed / content policy), we do not deliver that output to the user.
• Usage patterns: we monitor abnormal API volume and repeated policy failures on an account.

There is no public feed, so we do not run a post-publication gallery scan. Private outputs can still be reviewed when reported.

3.2 Human review

• User reports to hykflb@qq.com are reviewed by the ChimeraHub operator.
• Auto-blocked or repeatedly failing accounts are sampled for human confirmation, especially L1 (child safety, terrorism).
• Review is performed against this Policy. We keep records of automated refusals and human decisions for compliance review.

3.3 Periodic audits

• We review refused prompts, user reports, and upstream moderation errors at least weekly, and add terms to the blocklist when gaps appear.
• We do not currently commission an independent third-party audit.

3.4 Severity levels

Level	Description	Review	Action
L1 — Critical	Child safety, terrorism, WMD, non-consensual sexual deepfakes	Auto-block + human confirmation	Immediate ban; report to authorities where required
L2 — High	NSFW, gore, hate, clear IP theft, impersonation	Auto-flag + priority review	Remove content; suspend account
L3 — Medium	Other confirmed Policy breaches	Human review	Warning + removal; may restrict features
L4 — Low	Borderline or first-time minor issues	Report-triggered	Warning; require the user to stop

4. When and how to report

4.1 When to report

• Generated content or prompts that fall into any of the six prohibited categories.
• Critical violations involving child safety or terrorism.
• Deepfakes or other impersonation, or infringement of your copyright, trademark, or likeness.
• Attempts to jailbreak filters or to resell access for prohibited use.

4.2 Principles

• Confidentiality: we do not publish a reporter’s identity.
• Impartiality: reports are reviewed against this Policy, not against the reporter’s or target’s account standing.
• Anti-abuse: malicious or false reports are logged and may themselves lead to enforcement.

5. Reporting channels

Use the monitored channel below. Do not use social media as a substitute for a report.

• Report / safety / appeals: hykflb@qq.com
• In the email subject, write [Report] or [AUP], and include: your account email, approximate time, request or generation ID if you have one, a description of the content, and why it violates this Policy. Do not attach CSAM; describe it and we will preserve logs.
• Same mailbox for ordinary support. L1 reports are handled first.

6. Response timelines

Type	First reply	Decision target
L1 — Critical	Within 12 hours	Within 24 hours
L2 — High	Within 24 hours	Within 3 business days
L3 — Medium	Within 3 business days	Within 7 business days
L4 — Low	Within 5 business days	Within 15 business days

Workflow: received → triage (L1–L4) → automated evidence + human review → decision → enforcement → notice to the reporter when we have a valid address. If we need more time, we will say so and why.

7. Enforcement

Action	What we do	Typical use
1. Warning	Email notice; require you to stop	First L4
2. Content removal	Delete or withhold violating outputs we still control	Confirmed L3–L4
3. Feature restriction	Disable Studio, image routes, or lower quota	Repeat L4 or a single L3
4. Account suspension	Temporarily lock login and API keys	L2 or accumulated violations
5. Permanent ban	Close the account; unused credits are forfeited	L1 or repeat L2
6. Legal referral	Preserve logs and report to competent authorities	Criminal-level L1, especially CSAM

Credits spent on prohibited generations are not refunded. Unused credits may be forfeited on a permanent ban for L1 or repeat L2.

8. Appeals

You may appeal within 10 days of the enforcement notice by emailing hykflb@qq.com with subject [Appeal]. We will re-review the decision against this Policy and reply within 10 business days. If the original decision stands, we will say why.

9. Updates

Material changes will be posted on this page and, where we have an email, notified at least 15 days in advance. Continued use is acceptance.

10. Contact

• Safety, reports, and appeals: hykflb@qq.com
• Operator: ChimeraHub
• Website: https://api.chimerahub.org/

Last updated: 13 August 2026 · ChimeraHub · https://api.chimerahub.org/aup

ChimeraHub（“平台”“我们”）提供 AI API 接入与 Studio 图像生成。本《可接受使用政策 / AIGC 使用规范》（“本政策”）适用于经本平台生成、编辑、分发或存储的全部内容（含提示词）。所有用户均须遵守。本政策是服务条款第 10 条的细则。

Studio 生图底层模型为 GPT Image（gpt-image-2），经 image-auto 路由。文本及其他 API 线路使用 OpenAI、Anthropic、Google 的主流模型（如 GPT、Claude、Gemini 系列）。ChimeraHub 是独立产品，不是任何模型官方应用。输出均为 AI 生成。

1. 适用范围

• 文本：API 生成的文本、对话、代码等。
• 图像：Studio 与图像 API 生成或编辑的图像。
• 用户输入（提示词）：指令、参考图与参数。
• 用于产生上述内容的账号与 API 活动。

我们目前不提供公开作品广场或信息流。生成结果默认仅存在于请求账号中，除非你自行外传。

2. 内容标准与禁止类别

平台禁止生成以下六类内容，逐项列出，不以笼统表述替代：

1. 色情 / NSFW — 色情、淫秽或不适于工作场所的图像或文本，包括以性为目的的裸露、癖好内容与色情角色扮演。
2. 暴力 / 血腥 — 写实重伤、酷刑、肢解、现实暴力犯罪，或美化对人或动物造成严重伤害的内容。
3. 仇恨言论 — 基于种族、民族、国籍、宗教、性别、性取向、残疾等特征，攻击、非人化或煽动敌意的内容。
4. 儿童不安全内容（CSAM） — 任何涉及 17 岁及以下未成年人的性化、色情或剥削内容，包括虚构、绘画或 AI 生成的未成年人形象。零容忍。
5. 深度伪造 / 冒充他人 — 仿冒真实个人的 AI 音频、视频或图像，或故意误导来源与真实性的内容，包括未经同意的性深伪、明星或政治人物冒充。
6. 侵犯版权 / 商标 — 未经授权复制或攀附他人受保护的角色、作品、标识、品牌名称或其他商标。

2.1 绝对禁止 — 零容忍

• 儿童性剥削材料（CSAM）及任何涉及未成年人的性内容。
• 宣扬或协助恐怖主义、极端主义或大规模暴力的内容。
• 大规模杀伤性武器的技术说明。
• 煽动灭绝种族或民族仇恨的内容。
• 针对真实个人的非自愿性深伪内容。

2.2 其他禁止行为

• 试图绕过本平台或上游安全过滤的越狱提示词。
• 将 AI 生成内容伪装成人类创作以欺骗他人。
• 批量垃圾信息、钓鱼、恶意软件或网络攻击材料。
• 将内容标榜为执业医生、律师或金融顾问的专业意见。

即使有年龄验证，也不允许色情 / NSFW。请勿提交此类请求。

3. 平台内容审核机制

我们采用多层审核：输入与输出自动过滤、上游模型安全、举报人工复核，以及定期更新拦截词表。Studio 与图像 API 不是无审核的裸模型透传。

3.1 自动审核（实时）

• 输入过滤：提示词在转发前经过敏感词 / 拦截表检测。覆盖 NSFW、CSAM 相关、暴力与仇恨、深伪 / 冒充、受版权保护角色与品牌名称等，并随滥用模式更新。
• 上游安全：生图请求开启提供方审核（GPT Image 的 moderation: auto）。OpenAI、Anthropic、Google 对提示词与输出另有安全分类。被拒请求不会完成；Studio 会提示内容政策错误。
• 输出过滤：若上游或我方过滤器拒绝结果（如 content_moderation_failed），我们不会把该输出交给用户。
• 用量异常：监控异常 API 量与同一账号反复触发政策失败。

没有公开信息流，因此不做“发布后广场扫描”。私有产出被举报后仍会复核。

3.2 人工审核

• 发至 hykflb@qq.com 的举报由 ChimeraHub 运营者审核。
• 自动拦截或反复失败的账号会抽检，L1（儿童安全、恐怖主义）优先人工确认。
• 依据本政策审核，并保留自动拒绝与人工决定记录，以备合规查阅。

3.3 定期复盘

• 至少每周查看被拒提示词、用户举报与上游审核错误，发现缺口即补充拦截词。
• 目前未委托独立第三方审计。

3.4 严重等级

等级	说明	审核	处置
L1 危急	儿童安全、恐怖主义、大规模杀伤性武器、非自愿性深伪	自动拦截 + 人工确认	立即封禁；依法向主管机关报告
L2 高	NSFW、血腥、仇恨、明显侵权、冒充	自动标记 + 优先复核	删除内容；暂停账号
L3 中	其他已确认违规	人工审核	警告 + 删除；可限制功能
L4 低	临界或首次轻微问题	因举报触发	警告；要求立即停止

4. 举报原则

4.1 应当举报的情形

• 落入上述六类红线的生成内容或提示词。
• 涉及儿童安全或恐怖主义的危急违规。
• 深度伪造、冒充，或侵犯你的版权、商标、肖像。
• 试图越狱过滤，或将访问权转售用于违规用途。

4.2 处理原则

• 保密：不公开举报人身份。
• 公正：只按本政策审核，不按双方账号“地位”倾斜。
• 防滥用：恶意或虚假举报会被记录，并可能反制举报人。

5. 用户举报渠道

• 举报 / 安全 / 申诉：hykflb@qq.com（有人值守）
• 邮件标题请写 [Report] 或 [AUP]，并写明：账号邮箱、大致时间、请求/生成 ID（如有）、内容描述、以及违反本政策的哪一条。不要附上 CSAM 文件，文字说明即可，我们会保留日志。
• 普通客服也是此邮箱。L1 举报优先处理。

6. 响应时限

类型	首次回复	处理目标
L1 危急	12 小时内	24 小时内
L2 高	24 小时内	3 个工作日内
L3 中	3 个工作日内	7 个工作日内
L4 低	5 个工作日内	15 个工作日内

流程：收件 → 分级（L1–L4）→ 自动证据 + 人工审核 → 决定 → 处置 → 向有效地址的举报人反馈。若需延期，会说明原因。

7. 违规处置措施

措施	内容	适用
1. 警告	邮件告知，要求立即停止	首次 L4
2. 删除内容	删除或扣留我方仍控制的违规产出	已确认的 L3–L4
3. 限制功能	关闭 Studio / 图像线路或降低额度	反复 L4 或单次 L3
4. 暂停账号	暂时锁定登录与 API 密钥	L2 或累计违规
5. 永久封禁	关闭账号；未使用积分予以没收	L1 或反复 L2
6. 移送法律	保留日志并向主管机关报告	刑事级 L1，尤其是 CSAM

用于违规生成的已消耗积分不予退款。因 L1 或反复 L2 被永久封禁时，未使用积分可予没收。

8. 申诉

收到处置通知后 10 日内，可发邮至 hykflb@qq.com，标题 [Appeal]。我们将按本政策重新审核，并在 10 个工作日内答复。若维持原决定，会说明理由。

9. 更新

重大变更将发布于本页，并在我们持有邮箱时至少提前 15 天通知。继续使用即视为接受。

10. 联系方式

• 安全、举报与申诉：hykflb@qq.com
• 运营者：ChimeraHub
• 网站：https://api.chimerahub.org/

最后更新：2026 年 8 月 13 日 · ChimeraHub · https://api.chimerahub.org/aup