Why was my video generation blocked?

Sometimes a video request gets stopped before it ever reaches you. This page explains why that happens and how to avoid it next time.

Two layers of review

Your generation can get blocked at two different points, and it helps to know which is which:

Foundation model level. The underlying AI models that power video generation have their own built-in safety filters, set by the companies that build them. These filters are outside Cantina's control — we can't override them, and we don't always get a detailed reason when they trigger. If your prompt or image gets blocked here, you'll typically see a generic error rather than a specific policy citation.

Cantina level. On top of that, Cantina applies its own review for content that violates our Community Guidelines. This layer exists because being blocked at the foundation level isn't a guarantee something is actually safe — and because some things are against our rules even if a foundation model would technically generate them.

Both layers exist to keep Cantina legal, safe, and sustainable — not to police your imagination. We built this platform to let you create weird, funny, dark, and provocative stuff. The lines below exist because a small number of categories carry real legal and safety risk, for you and for the platform.

Common reasons generations get blocked

Explicit pornographic content

Cantina allows suggestive, romantic, and sexually-charged content, but fully explicit pornographic video — real or AI-generated — is not allowed. This includes:

• Explicit sexual acts or visible genitalia in video output
• Prompts specifically engineered to produce pornographic content
• Using someone's uploaded photo (yours or someone else's) as a base for sexual content without their involvement or consent

Where exactly "suggestive" ends and "explicit" begins can feel like a judgment call, and moderation errs toward caution here. If you're pushing right up against that line, expect more blocks.

Non-consensual intimate imagery (NCII)

Separate from explicit content generally, Cantina has zero tolerance for generating intimate or sexualized content of a real, identifiable person who hasn't consented to it. This includes:

• Uploading someone else's photo to generate nude, sexual, or intimate content of them
• "Undressing" or "nudify" style prompts applied to a real person's image, including public figures and celebrities
• Sexualized content of yourself generated to be shared or used against your own wishes (e.g., under coercion from another user)

This applies regardless of where the source photo came from — a public profile, a photo someone sent you, or a public figure's press photo don't count as consent. NCII is treated as a serious violation on its own, separate from and in addition to our general pornography rules, and can lead to immediate account action and, where required, a report to law enforcement or NCMEC.

Hate speech

Content that dehumanizes, demeans, or promotes violence against people based on race, ethnicity, national origin, religion, disability, gender, gender identity, sexual orientation, or similar protected characteristics will get blocked. This includes:

• Videos depicting slurs, hateful symbols, or imagery tied to hate movements
• Prompts asking bots or the generator to mock or degrade a group of people
• Content that frames real-world violence or atrocities against a group approvingly

Satire, dark humor, and edgy jokes are part of Cantina's culture — the line is punching at a protected group versus punching within a bit. A roast of a fictional character or a specific person who consented to being roasted is different from content targeting people for who they are.

Intellectual property

Video generations that closely reproduce copyrighted or trademarked material without rights to use it can get blocked. This includes:

• Recreating scenes, characters, or footage from copyrighted movies, shows, games, or music videos
• Using a real brand's logo, likeness, or trademarked characters in a way that implies endorsement or ownership
• Prompting for a "deepfake" style video of a celebrity or public figure that could be mistaken for real footage

Fan content and parody have more room to breathe than 1:1 reproductions, but the more your output looks like it could be mistaken for the original copyrighted work, the more likely it is to be flagged.

Content harmful to minors and children

This is the one category with zero tolerance and no gray area. Any content that sexualizes, exploits, or endangers anyone who is or appears to be under 18 is blocked outright, including:

• Any sexual or suggestive content involving a minor or a character/avatar that appears to be a minor, real or fictional
• Content depicting minors in violent, abusive, or exploitative scenarios
• Attempts to "age up" language around a clearly child-like image or character to get around filters

There is no version of "but it's just a drawing" or "the character is actually 800 years old" that makes this acceptable on Cantina. This category is enforced more strictly than any other, including for accounts otherwise in good standing.

Best practices to avoid getting blocked

• Keep sexual content suggestive, not explicit. Implied and off-screen is generally fine; depicted is where you'll run into trouble.
• Don't upload photos of real people for sexual or degrading content, including yourself, without a clear, consensual use in mind.
• Never use someone else's photo to generate nude, "undressed," or intimate content of them — this is treated as NCII regardless of intent, even if meant as a joke.
• Aim jokes at fictional bots, consenting participants, or public ideas — not at real people or protected groups.
• Use original characters or clearly transformed versions of existing IP rather than trying to recreate a copyrighted scene shot-for-shot.
• Avoid combining childlike features (age, body type, school settings, etc.) with anything sexual or violent, even as a joke or edge case.
• If a prompt gets blocked, don't just reword it to sneak past the filter. Repeated attempts to evade moderation on the same content can lead to account restrictions, even if any single attempt didn't succeed.

If you think a block was a mistake

Moderation isn't perfect, and legitimate content sometimes gets caught by mistake. If your generation was blocked and you believe it shouldn't have been, use the "Report an issue" option on the blocked result to send it for human review. Include the prompt and any source images used — that context helps reviewers move faster.

What happens if I keep getting blocked?

Occasional blocks are normal and not tracked against you. Repeated, deliberate attempts to generate content in the categories above — especially involving minors — can result in warnings, feature restrictions, or account suspension. See our Community Guidelines for the full policy.