Acceptable Use Policy

Serifu translates manga pages and uses AI models to read text and to clean or redraw the artwork behind it. This policy applies to everything you process with Serifu: the pages you upload for translation and every image or text Serifu produces from them. It is part of our Terms of Service.

1. Prohibited content

Do not use Serifu to process, create, edit or distribute any of the following.

1.1 Sexual or NSFW content

Pornographic or sexually explicit material, including explicit nudity, sexual acts, and content made for sexual gratification. Serifu is not for adult manga.

1.2 Violent or gory content

Content that depicts, promotes or glorifies extreme graphic violence, gore, torture or cruelty to people or animals, or that encourages self-harm or violence against others.

1.3 Hate speech

Content that attacks, dehumanises, harasses or incites violence or discrimination against people based on race, ethnicity, national origin, religion, disability, sex, gender identity, sexual orientation, age or similar characteristics, including extremist or terrorist propaganda.

1.4 Content that endangers children

Any sexual or sexualised depiction of minors (child sexual abuse material, CSAM), including drawn, animated, AI-generated or "fictional" depictions, and any content that grooms, exploits or endangers children.

1.5 Deepfakes and impersonation

Using Serifu to alter images so that they falsely depict a real person, to impersonate a person, creator, publisher or organisation, or to pass off altered pages as the original work or an official translation.

1.6 Copyright and trademark infringement

Translating or distributing work you do not own or have the rights to, removing credits, watermarks or copyright notices, or using other people's trademarks in a misleading way. Serifu is a tool for books you have bought or have permission to translate, for your own use.

1.7 Zero tolerance

The following always lead to immediate account termination and, where the law requires, a report to the authorities: child sexual abuse material; content supporting terrorism; instructions for making weapons capable of mass harm; incitement to genocide; non-consensual sexual deepfakes.

2. How we check content

• Input check (automated). Every page is scanned with the OpenAI Moderation API (omni-moderation, which checks images and text for sexual content, sexual content involving minors, violence and gore, hate, harassment and self-harm) before it is read, translated or redrawn. The extracted text and the translation are scanned too.
• Output check (automated). Every redrawn image is scanned with the OpenAI Moderation API before it is returned to you.
• Serifu's page check. Our reading model also checks each page for what the moderation API does not cover, such as deepfakes and impersonation of real people, and marks copyright notices, credits and watermarks so they are never erased.
• Provider safety systems. The models themselves apply their own safety filters: Google's on the reading and translation model (Gemini) and OpenAI's on the image model that redraws artwork.
• When something is flagged, the page is refused automatically: it is not processed further, no credit is charged, and the same page is refused again if it is sent again. The flagged page is not stored.
• Human review. Every report we receive is reviewed by the operator of Serifu, who decides on the action in section 3.
• Regular audits. We periodically test these checks on sample material and adjust them.

Pages are not stored after translation, including pages refused by these checks (see the Privacy Policy).

3. What happens when rules are broken

level	examples	action
L1 severe	child safety, terrorism, other zero-tolerance content	immediate permanent ban, remaining credits forfeited, report to the authorities
L2 high risk	sexual content, graphic violence, hate speech, deepfakes, repeated infringement	content blocked, account suspended; permanent ban for repeat violations
L3 medium risk	infringement or other violations without clear intent	content blocked, written warning
L4 low risk	minor or first-time issues	warning and a request to change how Serifu is used

We may also cancel pending translations and remove linked computers from an account under review. If you think we got it wrong, reply to our notice or email support@serifu.net and we will look again.

4. How to report

Email abuse@serifu.net. This inbox is monitored by a person. Please include what you are reporting, where you saw it, and any links or screenshots. You can also use the contact form.

We respond within these times:

level	first response	resolution
L1	2 hours	24 hours
L2	24 hours	3 business days
L3	3 business days	7 days
L4	5 business days	15 days

If a case takes longer, we tell the person who reported it. Copyright owners can send takedown or infringement notices to the same address.