Responsible AI Policy

1. What we do

• Prompt screening. We check prompts for sexual content before they reach a model. A blocked prompt is never sent to the model and costs you nothing.
• Model safety systems. Every model on Bolly also runs its developer’s own safety checks, which can refuse a request or withhold an output. You’re never charged credits for a refused run.
• Human review of reports. A person reviews every report we receive. Reports involving real people, and especially minors or intimate imagery, go to the front of the queue.
• Privacy by design. We don’t train AI models on your uploads or generations. Paid plans can keep generations private, and runs made through the MCP server on a paid plan are private by default.
• Provenance. Some models embed invisible watermarks or provenance data in what they generate (Google’s SynthID, for example). We don’t strip them, and you mustn’t either.
• Working with authorities. We report child sexual abuse material to the National Center for Missing & Exploited Children (NCMEC) and respond to valid legal requests.

2. Real people: consent and likeness

Mix, Face Swap, Lipsync, video avatars, headshots, Animate, Voice Changer and similar tools work with real people’s faces, bodies and voices. These rules apply to all of them:

• It’s you, or you have permission. Only upload someone’s photo, video or voice if it’s you, or if that person has clearly agreed to you using it this way. Permission to post a photo isn’t permission to put that person into a new video.
• Never minors. Don’t upload images, video or recordings of anyone under 18 to these tools, even with a parent’s permission.
• Public figures. Parody, satire and commentary about public figures are fine when they’re obviously not real. Don’t make a public figure appear to endorse a product, say something they never said, or take part in real events in a way that could mislead people.
• People who have died. Respect the rights of their families and estates. Don’t use a deceased person’s likeness or voice commercially without authorization.
• Voices. Don’t recreate or imitate a real person’s voice without their permission, and never use a voice to impersonate someone, deceive people or get past voice-based security checks.
• Private individuals. Never target a private person, whether a classmate, coworker, ex-partner or stranger, with content made to embarrass, harass or intimidate them.

If you appear in content on Bolly without your consent, tell us and we’ll take it down.

3. Content that’s never allowed

The first four categories below are zero tolerance. They lead to immediate account closure without a refund, and we report them to law enforcement where required.

• Child sexual exploitation (zero tolerance). Any content that sexualizes minors, including fictional, drawn, animated or “aged-up” depictions of people who look under 18, and any content that grooms or endangers children. We report it to NCMEC.
• Non-consensual intimate imagery (zero tolerance). Sexual, nude or intimate images or video of a real person, made or shared without their consent. This includes AI “undressing”, “nudify” edits and sexual deepfakes.
• Violent extremism (zero tolerance). Content that praises, promotes or recruits for terrorist or violent extremist groups or attacks.
• Threats and incitement (zero tolerance). Credible threats of violence, or content that encourages violence against people or groups.
• Sexual content. Bolly doesn’t allow pornography, nudity or sexually explicit content of anyone, real or fictional.
• Graphic violence. Gore meant to shock, or content that celebrates real-world violence or suffering. Fictional action and horror are fine.
• Hate. Content that attacks, dehumanizes or promotes hatred against people because of race, ethnicity, national origin, caste, religion, disability, sex, gender identity, sexual orientation, serious illness or immigration status.
• Harassment and bullying. Content that humiliates, threatens or targets a person, or exposes someone’s private information.
• Self-harm. Content that promotes or encourages suicide, self-harm or eating disorders.
• Illegal activity. Content that promotes or helps with illegal activity, such as selling drugs or weapons, human trafficking or other crimes.

4. Deception, fraud and impersonation

• Deceptive deepfakes. Don’t create realistic content showing a real person saying or doing something they didn’t and present it as real.
• Fake endorsements and reviews. Don’t use AI people or real people’s likenesses to fake testimonials, reviews or endorsements.
• Fraud and scams. No phishing, no impersonation of people, brands or government agencies, and no investment, romance or charity scams.
• Fake documents and evidence. No forged IDs, documents, receipts, certificates or records, and no fabricated evidence for legal, insurance, academic or any other proceedings.
• Identity checks. Never use Bolly output to get past identity verification, liveness tests, facial recognition or voice authentication.
• Fake news. Don’t make realistic fake news reports, emergencies, disasters or public-safety events, and don’t imitate real news brands or presenters to mislead people.

5. Elections and civic processes

• Don’t create content that could mislead people about how, when or where to vote, or about who can vote.
• Don’t make realistic content showing candidates, officials or election workers saying or doing things they didn’t, or faking their endorsement.
• Don’t use Bolly to discourage people from voting, or to impersonate campaigns, election officials or government bodies.
• Political satire and commentary are welcome. Label them clearly, and follow the political advertising and AI disclosure laws wherever you publish.

6. Other people’s work

• Only upload content you have the right to use, and don’t use Bolly to infringe copyrights, trademarks or other rights.
• Templates can contain clips from music videos, films and shows that belong to other people. Keep template remixes personal and non-commercial unless you have the rights holders’ permission.
• Don’t use brands, logos or characters in a way that suggests their owners sponsor or endorse you.
• Don’t pass off Outputs as the work of an artist, studio or creator who didn’t make them.

Rights holders can use our copyright complaint process.

7. Share AI content responsibly

• When you share realistic AI content, especially content showing real people or events that look real, say that it’s AI-generated. Use your platform’s AI labels (TikTok, YouTube, Instagram and others require them for realistic content), and follow any laws that require disclosure.
• Don’t remove, blur or alter watermarks, provenance data or AI labels added by Bolly, a model or a platform.
• In ads and sponsored posts, follow advertising rules: no fake testimonials, and every claim must be true.

8. Don’t work around our safeguards

• Don’t try to get past our filters or a model’s safety systems, for example by rewording blocked prompts, using coded language, misspellings or other languages, or splitting a request into pieces.
• Don’t use more than one account to avoid a ban or to collect extra free credits.
• Don’t use automation or the API to generate prohibited content at scale, and don’t help anyone else do any of these things.

Repeatedly trying to get around our safeguards can get your account suspended, even if nothing was generated.

9. How we enforce this policy

We use automated screening, model safety filters, reports from users and rights holders, and human review. When content breaks this policy, we may:

• block the request or remove the content;
• make a public generation private or remove it from Explore;
• limit features such as likeness tools or public sharing;
• suspend or close the account and cancel its credits; and
• report the conduct to NCMEC, law enforcement or other authorities where the law requires it or someone is at risk.

What we do depends on how serious the harm is, whether it looks intentional and the account’s history. The zero-tolerance categories always lead to immediate closure. When we act on your content or your account, we’ll normally tell you why. You can appeal within 30 days by replying to our notice or emailing support@bolly.ai, and a person will review your appeal.

10. Report a problem

Email abuse@bolly.ai with a link to the content (or its generation ID), what’s wrong, and how you’re connected to it (for example, you’re the person shown or you represent them). We review every report.

• Intimate images of you. If someone used Bolly to make or share sexual or intimate imagery of you without your consent, email abuse@bolly.ai with “Intimate image removal” in the subject. Include your name and contact details, your signature (typing your full name is enough), information that lets us find the content, and a short statement that you believe in good faith it was made or shared without your consent. We’ll remove it, and make reasonable efforts to remove any known copies on Bolly, within 48 hours of receiving a valid request.
• Your face or voice. If you appear in content on Bolly without your consent, we’ll review it and remove it wherever it breaks this policy.
• Copyright and trademarks. Use our copyright complaint process.
• Child safety. Report child sexual abuse material to us and to NCMEC’s CyberTipline at report.cybertip.org (outside the US, to your national hotline, such as the Internet Watch Foundation in the UK). Don’t download or share the material, not even to report it.
• Law enforcement. Send legal process to legal@bolly.ai.
• Emergencies. If someone is in immediate danger, contact your local emergency services first.

11. Changes to this policy

We update this policy as the technology, the law and the ways people misuse AI change. We’ll post updates here and change the date at the top of this page. Our Terms explain how we tell you about material changes.