Acceptable Use Policy — AI-generated content

H3 Max Studio · minimaxh3-max.org · last updated 4 September 2026

Purpose

H3 Max Studio (the platform, we) publishes this Acceptable Use Policy — AI-generated content for every prompt, upload, reference file, and video created at minimaxh3-max.org. This policy sets content standards, our content moderation process, how to report abuse, and how we enforce violations. Every user of text-to-video, image-to-video, and reference-to-video must follow it.

1. Scope

This policy covers all content forms on the studio:

• Text: prompts, instructions, expanded prompts, and on-screen copy in a clip.
• Images: first-frame, last-frame, and reference images you upload.
• Audio and video: reference clips you upload and AI-generated MiniMax H3 Max video with native stereo audio.
• User input (prompt): every instruction submitted before we call fal.

2. Content standards and prohibited categories（内容标准与禁止类别）

The platform explicitly prohibits using this service to generate the following six categories. Each item is listed separately and is enforced. Later sections add zero-tolerance rules and other abuse.

• 色情 / NSFW content.
• 暴力 / 血腥 content.
• 仇恨言论.
• 儿童不安全内容（CSAM）.
• 深度伪造 / 冒充他人（Deepfake）.
• 侵犯版权 / 商标 content.

2.1 Zero tolerance

The following is removed on discovery. We may report it to competent authorities:

• 儿童性剥削内容（CSAM） or any sexual content involving minors.
• Content that promotes or assists terrorism, violent extremism, or mass violence.
• Instructions for producing weapons of mass destruction.
• Incitement of genocide or ethnic hatred.
• Non-consensual sexual deepfakes of real people.

2.2 Restricted material

This is a video-generation studio. Adult / NSFW content is not offered even with age checks. Other restricted cases:

• Violence: only non-graphic fictional or educational action is allowed; realistic harm to identifiable people is not.
• Political content that could mislead viewers about real events or elections.
• Medical, legal, or financial advice presented as professional guidance.

2.3 Other prohibited conduct

These uses are also banned:

• Generating or spreading false information intended to deceive.
• Impersonating a real person, brand, or institution.
• Infringing intellectual property, publicity, or privacy rights.
• Presenting AI video as human-filmed footage in order to deceive.
• Bulk spam or fraudulent generation.
• Jailbreak prompts meant to bypass safety filters.

3. Content moderation（平台审核机制）

We use automatic filters, human review of reports, and sampling of recent jobs. We do not pass raw prompts to fal without a safety step.

3.1 Automatic review (real time). Input: every prompt is scanned with Waffo Prompt Sift (POST /v1/actions/verification/scan-prompt, semantic enforce) before we reserve credits or call fal. Generation continues only when action is allow. A block verdict shows a usage-guidelines message and does not create a job. A review or service-degraded verdict holds the request and asks the user to retry. Output: fal MiniMax H3 Max always runs enable_safety_checker. We keep job IDs, prompts, verdicts, and status in our logs.

3.2 Human review. Jobs that fail safety, and every report sent to [email protected], go to Trust & Safety. Reviewers look at the prompt, uploads, and output when available. We do not run a public gallery, so there is no separate post-publish crawl.

3.3 Sampling. When we investigate a report we also sample nearby jobs from the same account. We use those reviews to tighten block lists for NSFW terms, CSAM, copyrighted characters, violence, hate, self-harm, and deepfake / celebrity impersonation.

3.4 Severity. L1 severe: child safety or terrorism — automatic block plus human confirmation, then ban and report. L2 high risk: confirmed prohibited video — remove the clip and pause the account. L3 medium: other policy breaches — warning and removal. L4 low: first minor issue — warning and a required change.

4. Reporting principles

Report policy violations, child-safety or terrorism concerns, and deepfakes or IP claims that affect you. We keep reporter identity confidential, review reports independently, and record abusive false reports.

5. Reporting channels（用户举报渠道）

Email [email protected]. Put Report and the job ID in the subject when you have one. Include a URL, screenshot, or description of the clip. This mailbox is staffed. We do not operate a phone line or live chat. The same address is printed in the site footer.

6. Response times

L1 severe: first response within 2 hours; action within 24 hours. L2 high risk: first response within 24 hours; action within 3 business days. L3 medium: 3 business days to first response; 7 to finish. L4 low: 5 business days to first response; 15 to finish. If we need more time we say so. Flow: receive report → grade L1–L4 → review → decide → apply the action → notify when we have a working reply address.

7. Enforcement（违规处置机制）

We only list actions we can take on this studio:

• Content warning: written notice to change or delete a prompt or clip (first minor breach).
• Takedown: remove the video from history and stop downloads (confirmed L3–L4).
• Feature limit: pause new generations or lower remaining credits (repeat minor or one L3).
• Account suspension: temporary block pending review (L2 or repeated breaches).
• Permanent ban: close the account (L1 or repeated L2).
• Legal referral: report to law enforcement and cooperate on suspected illegal L1 cases, including CSAM.

8. Appeals

If you disagree with a decision, email [email protected] with Appeal and the job or account ID within 14 days. A reviewer who did not make the first decision re-checks the file. We aim to finish appeals in 10 business days.

9. Updates

Material changes to this policy will be posted on this page. We aim to give at least 15 days’ notice for non-emergency changes. Continued use after an update means you accept the new text.

10. Contact

Content safety, reports, appeals, and customer support: [email protected]. Operator: H3 Max Studio, minimaxh3-max.org. This policy is not legal advice.