AI Safety Policy

We keep AI on Pixeloop safe in three ways:

• Clear rules about what the tools may be used for, set out in this policy.
• Automatic safety checks on every generation request, before anything is made.
• People who review anything submitted to our public gallery, and who act on reports.

These rules apply to everyone who uses Pixeloop, whether you are signed in or using it as a guest.

• Image generation (Pixelsmith). Creates images from your written prompt and, if you add them, reference images you upload.
• Video generation (Motionsmith). Creates short videos from your written prompt. Some models can also add sound.
• Background removal. Removes the background from an image you choose.
• Image upscaler. Runs entirely on your own device. Your image is not sent to us or to any AI provider.

Our generation features are powered by established AI model providers, such as OpenAI, Google and fal. You choose the model before you generate.

You must not use Pixeloop’s AI tools to create, edit or share content that:

• sexualises children in any way, including drawn, animated or fictional images. We report this material to the authorities.
• is sexually explicit and shows a real person, or is an intimate image of anyone made without their consent;
• promotes terrorism, violent extremism, or serious violence against people or animals;
• harasses, threatens or demeans a person or group, including hateful content based on race, ethnicity, religion, caste, gender, sexual orientation, disability or any other protected characteristic;
• encourages suicide, self-harm or eating disorders;
• is made to deceive, such as fake news photos or videos, false evidence, forged documents or IDs, or material for scams and fraud;
• misleads people about elections or public life, such as fake images of candidates or false information about how to vote;
• infringes someone else’s copyright, trademark, privacy or publicity rights.

These are examples, not a complete list. Content that is illegal or clearly harmful is not allowed, even if it is not named here.

• Only upload photos you have the right to use. If a photo shows someone other than you, make sure you have their permission.
• Do not put a real person in a false or harmful situation. That means no images or videos that show someone in a sexual, violent or humiliating way, or doing or saying something they never did in a way that could fool people.
• Public figures. Creative and editorial work that features public figures, such as sports posters, fan art or parody, is allowed. It must not be passed off as a real photo or video, suggest that the person endorses something when they do not, or damage their reputation.

• Every prompt you send to Pixelsmith or Motionsmith, and every reference image you add, is checked by the safety systems of the provider that runs the model. If the request breaks their rules, it is blocked and nothing is made.
• When a request is blocked, we tell you. A blocked request does not use your credits. You can change your wording or your reference image and try again.
• Automatic filters are not perfect. They can sometimes block a harmless request, and they can sometimes miss something harmful. That is why the rules in this policy still apply to you, even when a request gets through.
• Deliberately trying to get around a safety filter, for example with coded words or by splitting a prompt into pieces, is a breach of this policy.

• The images and videos you generate are private to your account unless you choose to share them.
• Share links. If you create a share link, anyone who has that link can view what you shared. You are responsible for what you share and who you share it with.
• Community gallery. When you submit an image to our inspiration gallery, a member of our team reviews it by hand before anyone else can see it. We reject submissions that break this policy, and we may remove an image later if we find a problem with it.

• Images and videos made with Pixelsmith and Motionsmith are created by AI. If you publish one where people could mistake it for a real photo, real footage or a real event, make it clear that it was made with AI, and follow the rules of the platform you post it on.
• Some of our providers may add invisible markers or metadata that identify an image or video as AI-generated. Do not remove or tamper with them.
• AI can get things wrong. It can misspell text, add odd details or strange movement, or show things that look real but are not. Always check your result before you use it.

• We do not use your prompts, uploads, or generated images and videos to train AI models.
• We send only what is needed to create your result to the provider that runs the model you chose. Each provider handles that data under its own terms, which may allow it to keep the data for a limited time to detect abuse.
• Saved images and videos are kept for 30 days and then deleted automatically. You can delete any of them sooner. When you do, the image is removed straight away and the rest of the record is erased within 7 days.

For full details on how we handle personal data, see our Privacy Policy.

If you find content made with Pixeloop that you believe breaks this policy, or you find an image or video of yourself that was made without your consent, please tell us. Email hello@pixeloop.app with the subject “AI safety report”, and include a link or screenshot and a short description of the problem.

• We read every report and act on it as quickly as we can.
• Reports that involve children or a risk of immediate harm are handled first.
• We keep the identity of the person who made a report private.

Depending on how serious the breach is, we may:

• block the request or remove the content;
• send you a warning;
• limit your access to AI features;
• suspend or close your account, without notice for serious breaches; and
• report illegal content to law enforcement, as we always do for child sexual abuse material.

If you believe we made a mistake, email us and we will look at the decision again.

AI technology and the laws around it are changing quickly, and this policy will change with them. When we make an important change, we will update the date at the top and, where it affects you, tell you in the app or by email.

Questions about this policy: hello@pixeloop.app.