Acceptable use policy

Forbidden content

1. Sexual content and nudity. No nudity, pornography or sexual content of any kind.
2. Content that endangers children. Nothing that sexualises, endangers or exploits anyone under 18, and no images of real children.
3. Violence and gore. No graphic violence, gore, self-harm, or images that threaten or glorify harm to real people or groups.
4. Hate and harassment. No hateful symbols, slurs, or images that demean people for their race, ethnicity, religion, gender, sexual orientation, disability or nationality, and nothing meant to bully or harass someone.
5. Deepfakes and impersonation. No images of real people (celebrities, politicians or private persons) without their consent, no edits that put a real person into a scene they were not in or show them doing or saying something they did not, and no images meant to pass as real photos of real events. Face swaps, “undressing” or “nudify” edits and intimate scenes (kissing, hugging, in bed) made from photos of real people are forbidden even when the people in them agree.
6. Copyright and trademarks. No copies of other people’s artwork, characters, logos or brands that you do not have the right to use.

Also forbidden: anything illegal where you live or in the European Union or the United States, fraud and scams, fake documents or evidence, and spam or automated mass generation.

Images you upload to edit

Upload only images you made or have the right to use. A photo of a real person must be a photo of you, or you must have permission from everyone who can be recognised in it. Do not upload photos of other people taken from social media or the web, and do not upload photos of children.

These edits are never allowed, even if the people in the photo agree:

• Face swaps: putting one person’s face on another person’s body or photo, or turning someone into another real person.
• “Undressing” or “nudify” edits: making a real person look nude or sexualised.
• Intimate scenes: putting real people into a kiss, a hug, a bed or any other intimate scene they were not in.
• Fabricated actions: showing a real person doing or saying something they did not do or say.

How we check

• Prompts. Before anything is sent to the model, our server checks every prompt against a list of terms for each of the six kinds of forbidden content, and for edits of an uploaded photo, against terms for intimate scenes. We update the list as we see new ways of asking. A prompt that matches is refused, and nothing is charged.
• Uploaded images. Before an image you upload is sent to the model, an automated image check looks for nudity and sexual content, children, and graphic violence. A refused image is not sent, and nothing is charged.
• Finished images. Every image the model makes is checked the same way before it is shown to you. An image that fails is never shown and its link is not kept.
• If a check cannot give an answer (for example, because the checking service is down), we treat that as a refusal: no image is made or shown.
• The model runs on Black Forest Labs’ API with its own safety filter, which also turns some requests and results down.
• Every refusal is logged with the stage (prompt, uploaded image or finished image), the kind of content, which check refused it, the time, and the wallet code or a hashed visitor ID. Images are never stored in the log. The prompt is deleted from the log after 30 days and the log entry after 365 days. Prompts of images that were made are kept for 30 days, so that we can look into reports.
• Our checks are automated and can be wrong in both directions. We review reports by hand.

What happens when the rules are broken

• After 5 refusals in one day (prompts, uploaded images or finished images), free images stop for that visitor until the next day.
• A wallet code with 10 refusals is blocked: it can no longer make images. We may also block a wallet code after a report. Images used to break these rules are not refunded.
• We report child sexual abuse material and other illegal content to the authorities.

Sharing images

When you post an image made here, say that it was made with AI where that could matter to the people who see it, and follow the rules of the platform you post on.

Report a problem

If an image made on this site shows you without your consent, copies your work, or breaks these rules in another way, write to support@flux3klein.org with a description and, if you have it, where you saw it. We reply within 3 business days. If you think one of our checks refused your request by mistake, write to the same address.