Published: Jul 24, 2026
Updated: Jul 24, 2026
Operator: AIYONG TECHNOLOGY CO., LIMITED
Related pages: Terms of Service · Report Content · Privacy Policy

This page describes our content filtering methods, review workflow, user reporting channel, human review process, content takedown rules, and account enforcement. It reflects the controls currently implemented on aimakesong.com.

1. Scope of Covered Features

These rules apply to user prompts, uploads, edits, and outputs across our platform, including but not limited to:

AI music generation (text-to-song, lyrics-to-song, hum-to-song, and related tools)
AI lyrics / text generation
AI image generation and editing (text-to-image, image-to-image)
AI video generation (including Seedance / multimodal video flows)
Audio tools (vocal remover, stem splitter, audio cutter, audio-to-MIDI, and similar)
Music video / lyric video / singing photo style tools
Any upload, share, history, or download surfaces connected to the above

2. Prohibited Content Categories

We prohibit creating, requesting, uploading, processing, distributing, or attempting to bypass filters for the following categories:

Category	Examples (non-exhaustive)	Default action
Sexual / pornographic content	NSFW, pornography, sexually explicit imagery or audio, fetish content	Block / refuse generation; remove if discovered
Violence & graphic harm	Graphic gore, glorification of real-world violence, terrorist propaganda	Block / refuse; escalate for severe cases
Illegal activity	Drug trafficking instructions, fraud guides, weapons misuse guidance, other clearly illegal facilitation	Block / refuse; escalate
Infringement	Uploading or requesting copyrighted lyrics/melodies/voices/likenesses without rights; trademark abuse	Block where detected; remove on notice; user remains liable
Deepfakes & non-consensual synthetic media	Face-swap, celebrity likeness impersonation, realistic fake identity media intended to deceive	Pre-generation block (keyword/pattern guard + moderation); remove if discovered
Minors & child exploitation	Any sexual or exploitative content involving minors (fictional or real); grooming content	Zero tolerance: block, remove, suspend/ban, retain evidence as required by law

We also prohibit attempts to evade filters (obfuscated prompts, split instructions, adversarial wording).

3. How Filtering Works (Aligned to Current Product)

3.1 Policy layer (always on)

Terms of Service prohibit illegal, harmful, infringing, NSFW, deepfake, and minor-related abuse.
Users must confirm they have rights to uploaded materials and that their use complies with law.

3.2 Automated prompt moderation (AI image / video)

For AI image and video generation routes (including text-to-image, image-to-image, and Seedance-related generation), we apply fail-closed checks before generation:

Deepfake / real-person synthetic guard — keyword and pattern matching against face-swap, deepfake, celebrity likeness, and related intent (including common English and Chinese phrases). Matched prompts are rejected with an error and are not sent to the model.
Third-party prompt moderation API (Creem Moderation) — prompts are evaluated with decisions:
- allow → generation may proceed
- deny / flag → generation is refused; user is asked to revise
- API failure / timeout / misconfiguration → generation is refused (fail closed)

3.3 Model / provider safety

Upstream AI providers maintain their own safety policies and may independently refuse or filter illegal or high-risk requests. Platform refusal and provider refusal can both stop a job.

3.4 Coverage note (honest capability statement)

Strongest automated pre-checks today are on image/video prompt paths listed above.
Music and audio tools rely on Terms prohibitions, provider-side controls where available, post-hoc review from reports, and account enforcement.
We continuously expand automated coverage; absence of a pre-check does not mean prohibited content is allowed.

4. User Reporting Channel

Anyone can report suspected violating content or accounts:

Report page: https://www.aimakesong.com/report-content
Email: [email protected] (subject line: Content Report)

Please include:

Your contact email
Content URL, share link, task/history ID, or screenshot
Category (sexual, violence, illegal, infringement, deepfake, minors, other)
Short description of the issue

5. Review, Takedown, and Account Actions

5.1 Intake & triage

Priority	Typical cases	First response target
P0 — Critical	Minors exploitation, imminent real-world harm, clear illegal facilitation	Acknowledge within 24 hours; begin removal/suspend immediately where verified
P1 — High	Deepfake impersonation, NSFW distribution, severe violence, clear IP theft with evidence	Acknowledge within 1–2 business days
P2 — Standard	Other policy violations, unclear edge cases	Acknowledge within 3 business days

5.2 Human review

Reports are reviewed by support / trust & safety staff (not only automated systems).
Reviewers check account history, prompts/uploads where available, generated outputs, and prior warnings.
For infringement claims, we may request additional evidence of rights ownership.

5.3 Content actions

When a violation is confirmed, we may:

Refuse or cancel generation jobs
Delete or disable access to generated outputs / uploads / share links
Remove content from history or public surfaces
Preserve limited records needed for abuse prevention, billing disputes, or legal compliance

5.4 Account enforcement ladder

Stage	When used	Effect
Warning	First mild/borderline violation	Notice via email or in-product message
Feature restriction	Repeated policy abuse	Temporary block of generation/upload for specific tools
Suspension	Serious or repeated violations	Temporary account lock
Permanent ban	Minors exploitation, severe illegal use, repeated deepfake/NSFW abuse, fraud	Permanent termination; credits/subscription may be forfeited without refund where Terms allow

We may skip steps for severe violations.

6. Evidence Retention

Content reports, related ticket notes, and enforcement decisions are retained as needed for safety operations (typically up to 2 years, unless law requires longer).
Payment records follow our Refund / Privacy retention rules (legal/tax retention may be longer).
We do not publicly disclose investigation details that would help evade filters.

7. Appeals

If you believe content or an account was actioned in error, email [email protected] with subject Moderation Appeal, your account email, and any supporting evidence. We aim to respond within 5 business days.

8. Contact

AIYONG TECHNOLOGY CO., LIMITED
RM Al, 11/F, WINNER BUILDING, 36 MAN YUE STREET, HUNG HOM HONG KONG
Email: [email protected]
Website: https://www.aimakesong.com

Content Moderation & Safety Policy