What gets blocked when generating AI images and videos on ACE AI Spark, why, and what to try when a request is rejected.
Home › Help Center › Troubleshooting › Content guidelines for AI generation
# Content guidelines for AI generation Every image, video, and voice generation on ACE AI Spark passes through automated safety filters before it's produced. Most of the time you'll never notice these filters — they only kick in when a request crosses a clear line. If your generation came back with **"the input image violates the usage guidelines"** or a similar message, this page explains what that means and what to do next. ## What gets blocked The safety filters reject requests in a small number of categories. The most common reasons: ### Real people's faces Photos of identifiable real people — celebrities, politicians, public figures, or even friends and family — are blocked when used as input for video or image generation. This is to prevent impersonation and deepfake misuse. **What to do instead:** Use AI-generated faces from our avatar library, your own commissioned likeness, or stylized illustrations. ### Children and minors Any image that appears to depict a minor will be blocked regardless of context. ### Violence, gore, or weapons in realistic settings Action-movie style violence, weapons of war, or graphic injury imagery is blocked. Stylized fantasy combat (e.g. a knight in armor) is generally allowed. ### Sexually explicit or suggestive content Nudity, sexually suggestive poses, and adult content are blocked across the board. ### Copyrighted characters Recognizable cartoon characters, movie franchises, video-game characters, and similar IP cannot be regenerated. Original characters inspired by a genre are fine. ### Hate symbols and extremist imagery Symbols, flags, and slogans associated with hate groups or extremist ideologies are blocked. ### Identifying brand logos used commercially Generating content that uses identifying brand logos in a way that could be mistaken for an official endorsement may be blocked. ## What to do if your request was blocked 1. **Rephrase or rebuild the prompt.** Sometimes a single word triggers the filter even though the intent was innocent. Re-describing the same idea in different words often works. 2. **Swap the input image.** If you uploaded a reference image, try one that's clearly stylized, illustrated, or generated rather than a real photograph. 3. **Use a different scene description.** Move the action to a fantasy or abstract setting — the filter is much more permissive there. 4. **Check what you're asking for.** If the request genuinely involves a real person's face or any of the categories above, the filter is doing its job. There's no workaround. ## What about refunds? When a generation fails because of a content-policy block, **your credits are automatically refunded** within 30 minutes. You don't need to contact support — the refund goes back to your balance on its own. You'll see a notification in the bell icon once the refund processes. ## What about false positives? The filters aren't perfect. If you believe your request was blocked unfairly — for example, a stylized illustration that was treated as a real photo — reach out via the [Contact form](/contact) and we'll take a look. Include the prompt you used and a description of what you were trying to make. ## Why we have these rules The underlying generation systems we use require these rules as a condition of access. Beyond that, they're the right call for the platform — we want ACE AI Spark to be a tool serious people use for serious work, and that means keeping it free of the kinds of content that would erode trust in everything else generated here. If you have a use case you think we should support that doesn't fit the current rules, we want to hear about it — that's how the guidelines evolve.Help Center · Learn · Privacy Policy · Terms of Service · Contact