Content Policy

Every prompt is screened before it reaches a model, and every generated image is screened before it can appear anywhere public. This page is the short version of what that screen looks for.

Everything generated here is AI-generated. LevelShaper is not affiliated with, endorsed by, or sponsored by Roblox Corporation; content you publish on Roblox is also subject to Roblox's own Community Standards and moderation.

What you can make

Roblox games are for a young audience, and we hold generated content to that standard. Within it, the bar is wide:

• Obbies, tycoons, simulators, adventure maps, racing, building and social hangouts.
• Cartoon combat, lasers, swords, monsters and other game-style action.
• Original characters, worlds, items, thumbnails, icons, GUI and scripts.

What is not allowed

A prompt is rejected — before anything is charged — and a result is removed after the fact when it falls into any of these categories.

• Sexual content. Any sexual or sexually suggestive content, in any form.
• Anything that sexualises or endangers minors. Zero tolerance, no exceptions, and confirmed cases are reported to the relevant authorities.
• Graphic violence and gore, and content that glorifies real violence, real atrocities, self-harm or suicide.
• Hate and extremism. Hate symbols, extremist insignia, terrorist content, and content that attacks people for who they are.
• Real people. A real person's likeness or name used for harassment, impersonation, false endorsement, or without their consent.
• Scams and account theft. Fake "free Robux" generators, phishing pages, cookie loggers, or anything designed to trick players out of their accounts or money.
• Malicious scripts. Backdoors, exploits, cheats against other players' experiences, or code intended to break Roblox's rules.
• Copyright and trademark infringement. Someone else's characters, logos, brands or games reproduced as-is. Inspiration is fine; a copy is not.
• Bypass attempts — wording designed to route around the screen, including obfuscation, encoding and repeated rephrasing of a rejected request.

How moderation works

1. Automated prompt screening, before generation. Every prompt goes through a third-party content-safety check before any credit is charged. The screen is fail-closed: if it is unreachable or returns something we do not understand, the request is refused rather than allowed through. A rejected prompt costs nothing.
2. Automated image screening. Generated images are classified by an automated safety classifier before they can appear in a public gallery.
3. Human review of every report. A person reads every report we receive and decides what happens to the content.

The screen is automated and imperfect in both directions: it rejects things it should not, and things slip through. That is what reporting and appeals are for. Passing our screen does not guarantee that Roblox will accept the content.

Enforcement

• Prompt rejected — the generation never runs and nothing is charged.
• Content removed — we take it down, refund the credits on request, and tell you which rule it broke.
• Account suspended — for repeated violations or a single serious one.
• Permanent ban and referral — for content that sexualises minors, terrorist content and other zero-tolerance categories.

Reporting content

• Policy violations: support@levelshaper.com
• Copyright and trademark: dmca@levelshaper.com

Include the page URL and what is wrong with it. We reply within two business days, remove anything that violates this policy, and keep the reporter's identity confidential.

Appeals

If a prompt was rejected or content was removed and you think we got it wrong, email support@levelshaper.com within 14 days with the link or the rejection message. A person reviews it and replies within five business days. If we agree, the content is restored and any credits charged are returned.

Changes

We update this page when the screen changes. The date at the top is the last revision.