Acceptable Use Policy

A PrankLab prank should make the person it is about laugh once they find out. If it would not — if the point is to humiliate them, frighten them, mislead someone about them, or make money off them — it is not a prank and this is not the tool for it.

Everything below follows from that. This policy is part of our Terms of Service, and breaking it can cost you your account and your remaining credits.

You may upload a photo of yourself, or of an adult who has agreed to appear in an AI-generated prank. That is the whole permitted set.

Permission has to be real. It means the person knows their image will be altered by AI and that the result will be shared. It is not implied by them being your friend, your partner, your sibling, your colleague, or by the photo being on their public social media. If you would be uneasy asking them, you already know the answer.

Permission can be withdrawn. If someone tells you they want it gone, delete the upload and the result, and stop sharing it.

Do not use PrankLab to create, upload, generate, store or share anything in these categories. This list is not exhaustive, and the spirit of it counts as much as the letter.

People who cannot consent

• Any image of a person under 18, in any context, for any reason. This is an immediate and permanent ban, and we report child sexual abuse material to the National Center for Missing & Exploited Children and to law enforcement.
• Photos of anyone who has not given permission, including strangers, people photographed in public without their knowledge, and people who have refused.
• Photos of someone who has asked you to stop, or who has blocked you.

Sexual and intimate content

• Nudity, sexual acts, or sexually suggestive depictions of any identifiable person.
• Non-consensual intimate imagery of any kind, including undressing, body modification of a sexual nature, or placing someone in a sexual scenario. This is illegal in most places and is an immediate permanent ban.

Harm to a person

• Harassment, bullying, stalking, or a campaign of repeated unwanted content about someone.
• Threats of violence, or depictions of a real person being injured, killed, assaulted or in genuine distress.
• Content that degrades or attacks a person or group based on race, ethnicity, national origin, religion, disability, disease, age, sex, gender identity, or sexual orientation.
• Depicting someone committing a crime, using drugs, being arrested, or in any scenario that could realistically damage their reputation, employment, immigration status, safety, or family relationships.
• Content intended to cause panic, such as a fake accident, arrest, death, medical emergency, or disappearance involving a real person.

Deception and fraud

• Anything presented as a real event, real footage, real evidence, or a real news report to someone not in on the joke.
• Fake evidence for a court, an insurer, an employer, a landlord, a school, an immigration authority, or the police.
• Impersonating a police officer, government official, judge, doctor, journalist, or emergency service.
• Impersonating a real company, brand, news organisation or public institution, or using their logos and identity to make something look official.
• Scams, phishing, catfishing, romance fraud, fake dating or social profiles, extortion, or blackmail.
• Content about elections, voting, public health, disasters, or ongoing emergencies that could be mistaken for real information.
• Fake endorsements, testimonials, reviews, or advertising claims, including making it look like someone endorses a product.

Official-looking documents and panic formats

• Identity or travel documents — ID cards, driving licences, passports, visas, residence permits, student or employee badges.
• Financial or employment paperwork — bank statements, transfer confirmations, invoices, pay stubs, tax documents, insurance cards, diplomas, certificates, or professional licences.
• Tickets or passes with real monetary or access value — real lottery or scratch formats, event tickets, boarding passes, transit passes.
• Law-enforcement or government notices — parking tickets, citations, warrants, summonses, wanted posters, mugshots, immigration or benefits notices.
• Emergency or public-safety formats — fake 911/999/112 screens, Amber Alerts, weather or evacuation broadcasts, dispatch messages.
• Fear or harm scenarios — realistic intruders or break-ins, weapons pointed at people, death or funeral notices, serious injury, medical test results, scans, diagnoses, or prescriptions.
• Pixel-accurate clones of real apps (messaging, social, banking, email) or login, checkout, or “verify your account” flows for any real service.

Legal and technical limits

• Content that infringes someone's copyright, trade mark, publicity rights or privacy rights.
• Anything illegal where you live or where the person in the photo lives.
• Editing a result to strip provenance metadata, or otherwise disguising that it was generated by PrankLab.
• Scraping, automated access, reverse engineering, evading rate limits or content filters, sharing account credentials, or reselling access to the Service.
• Uploading malware, or using the Service to attack or overload our systems or anyone else's.

Satire of public figures is a long and protected tradition, and some of our templates exist for it. Obvious parody involving a celebrity or politician — an absurd tabloid cover, a ridiculous tour poster — is generally fine.

It stops being fine when it stops being obviously a joke. Do not make content about a public figure that a reasonable person could take as real, that puts words or actions on them they could plausibly have said or done, that is sexual, that depicts them committing a crime, or that touches an election or a public emergency. Commercial use of a public figure's likeness — using it to sell or promote anything — is never permitted.

Public figures have publicity and reputation rights too, and being famous is not consent.

PrankLab labels results as AI-generated in the product. Sensitive template categories also carry a visible watermark that cannot be removed. Everyday comedy templates may export without a visible watermark on paid plans. Every export embeds provenance metadata (for example C2PA / IPTC). Do not strip that metadata.

When you post anything generated here to a public platform, say that it is AI-generated. Most platforms now require this in their own rules, and in the European Union, Article 50 of the AI Act requires people who publish deepfake content to disclose that it is artificially generated. Where that law applies to you, complying with it is your responsibility, not ours.

Within a private group chat, the equivalent duty is simpler: tell them before it goes too far.

Email legal@pranklab.io. You do not need a PrankLab account to report something, and you do not need to be the person affected.

Tell us what you saw, where you saw it, and — if you have it — a link or a copy. If you are reporting an image of yourself made without your permission, say so and we will prioritise it. Our full process, including copyright notices and counter-notices, is in the Copyright, Likeness & Takedown Policy.

If someone is in immediate danger, contact your local emergency services first. We are not an emergency service.

We use automated filters and human review. Neither is perfect, so we also depend on reports.

We may also block generations before they run when a request clearly aims at something on this page. A blocked generation does not cost a credit.

Automated moderation makes mistakes and so do we. If your content or account was actioned and you believe it was wrong, email legal@pranklab.io within 30 days with your account email and what was removed. A person who was not involved in the original decision reviews it, and we aim to answer within 10 business days.

If we reverse a decision, we restore the content and any credits that were consumed, and we remove the strike from your account record.