Content rules, filters and safety

What Dreamina refuses to generate, why ordinary prompts sometimes trip the filter, and where the line sits between a blocked render and an account problem.

Last reviewed 13 August 2026

Short answer

What is not allowed on Dreamina?

Sexual content, graphic violence, content involving minors in any unsafe context, recognisable public figures, hate speech, and material designed to deceive — all filtered. The filter runs twice: once on your prompt before anything is spent, and again on the output before it reaches you. Most blocks people encounter are false positives from pattern-matching, not genuine violations, and rephrasing the one triggering word usually resolves them.

Boundaries

What the filter blocks

Restrictions fall into three bands, and they carry very different consequences.

Category	Treatment	Consequence
Sexual or adult content	Hard block, prompt and output	Repeated attempts escalate to the account
Content involving minors unsafely	Hard block, zero tolerance	Immediate account action
Graphic violence or gore	Hard block	Repeated attempts flagged
Recognisable public figures	Blocked or heavily degraded	Usually just a failed render
Hate speech and harassment	Hard block	Escalates to the account
Deceptive or impersonating material	Blocked where detected	Escalates if deliberate
Brand logos and trademarks	Inconsistently blocked	Your legal exposure, not theirs
Medical or anatomical terms	False positives common	None — rephrase and continue

The last two rows are where most confusion sits: one is permitted by the tool but risky for you, the other is blocked by the tool but entirely legitimate.

A blocked prompt is not a strike

Tripping the filter once, or fifty times by accident, is not an account problem. What escalates is a persistent pattern of attempts against the hard-blocked categories. If your architectural renders keep getting caught by a keyword, you are not in any trouble.

Practical

Getting a legitimate prompt past the filter

This is about clearing false positives, not circumventing the rules — the hard categories stay blocked whatever you write.

1
Identify the single triggering word rather than rewriting everything. Remove terms one at a time until it runs; the last one you removed is the culprit.
2
Replace clinical or anatomical vocabulary with plain description. "Chest" reads differently to a filter than the clinical term, with identical meaning to the model.
3
Add context that disambiguates. "A historical battlefield painting in the style of a museum oil" clears filters that a bare violence-adjacent noun does not.
4
Describe roles instead of naming people. "A 1960s American president at a podium" generates; naming him does not.
5
For brands, describe the design rather than the mark. You avoid both the filter and the trademark problem in one move.

Licensing

What you own, and what you do not

On a paid plan you receive a commercial licence for what you generate. That licence covers Dreamina's own claim on the output — it does not and cannot grant rights in third-party material the model reproduces. If your prompt yields something recognisably derived from a protected character, logo or living person's likeness, the commercial licence does not protect you from the rights holder.

Free-tier output is personal use only. The watermark makes that restriction visible, but the restriction exists whether or not the mark is present — which is the part most watermark-removal advice quietly skips.

Uploading photographs of other people

Reference images of people who have not consented are the highest-risk thing most users do without thinking about it. It is a data-protection question as much as a policy one, and it applies to friends and colleagues just as much as to strangers.

FAQ

Policy questions

Does Dreamina allow NSFW or adult content?

No. Sexual content is filtered at the prompt stage and again on output, and this is one of the few areas where the filter is deliberately aggressive rather than approximate. Repeatedly attempting it is treated as a policy issue, not a failed generation.

Is Dreamina AI uncensored?

It is not, and there is no setting that makes it so. Searches for an uncensored mode generally lead to unofficial sites that impersonate Dreamina — those are worth avoiding entirely, since they typically want your CapCut credentials.

Why was my perfectly ordinary prompt blocked?

Filters match patterns, not intent, so ordinary words collide with restricted ones — anatomical terms in a medical context, violence in a historical one, brand names in a design one. Rephrasing the specific triggering word usually clears it without changing your meaning.

Can I generate images of real people or celebrities?

Recognisable public figures are restricted, and generating identifiable real people without consent creates legal exposure regardless of what the tool permits. Uploading a photograph of someone else as a reference is the higher-risk version of this.

Do I lose credits when a generation is blocked?

A prompt rejected before it runs costs nothing. One blocked at the output stage has already consumed compute and is not always refunded automatically — if it happens repeatedly, that is worth raising with support.

Can I use Dreamina output commercially?

Only on a paid plan. Free-tier output is personal use only and carries a watermark. A paid plan grants a commercial licence for what you generate, though that licence does not override third-party rights in anything recognisable your prompt reproduces.

Keep reading

The watermark

Why free-tier output is marked, and the one supported way to export clean.

Plans and licensing

Which tier grants commercial rights.

Prompt guide

Writing prompts that are specific without tripping filters.

What Dreamina is

Who runs it and how your data is handled.

This page summarises publicly documented policy for orientation. It is not legal advice, and Dreamina Hub is not affiliated with Dreamina. The operative terms are the ones published on the official site, which take precedence over anything written here.