Back to home

Content & Safety

The AI models RoomMaker AI uses, what we will and won’t generate, how every prompt, photo and result is checked, and how to report a problem.

Last updated: 2026-10-09

Built for homes and spaces

RoomMaker AI is a design tool for rooms, homes and gardens, and everything it generates has to be appropriate for a general audience. The full rules are in our Acceptable Use Policy.

AI models we use

Every result on RoomMaker AI is generated by AI and is a concept visualization, not a photo of a real space.

  • Image generation — GPT Image 2 by OpenAI powers every image tool: room redesign, renovation previews, furniture staging, floor plan and 3D concepts, exterior and garden designs, listing photos and image edits made by the RoomMaker Agent. The quality tiers you choose — Basic 1K, Pro 2K and Ultra 4K — are output resolutions of this same model.
  • RoomMaker Agent — the assistant uses the DeepSeek V4 Flash language model to understand your request and decide which design tool to run. Images it creates come from GPT Image 2.
  • Safety review of prompts, photos and results — Gemini Flash by Google.
  • Text-only floor plan concepts — drawn by our own layout code, without an AI model.

Every image request runs through the same safety checks, whichever tool or quality tier you choose. We access these models through APIs. RoomMaker AI is an independent product and is not affiliated with, sponsored or endorsed by OpenAI, DeepSeek or Google.

What we don’t generate

  • Nudity, sexual or suggestive content of any kind.
  • Anything that sexualizes or endangers children.
  • Gore, graphic violence or self-harm imagery.
  • Hate symbols, extremist insignia or slogans.
  • The likeness of real, identifiable people, or deepfakes.
  • Copyrighted characters, brand logos or trademarks as the subject of a design.

If a request crosses one of these lines we say so plainly. Requests blocked before generation never use your credits.

How we check

  1. Before generation — every prompt is screened on our servers by a blocklist we keep updating, then by a safety model that reads what the request really asks for. Messages to the RoomMaker Agent are screened against the same blocklist.
  2. Before the image model sees your upload — photos, sketches and plans are checked by an image safety model.
  3. Before display — every generated image is checked by an image safety model before it is shown or saved. If it fails, it is discarded and your credits come back.
  4. If the checks are unavailable — requests are blocked rather than let through.
  5. Records — blocked requests and reports are logged and reviewed by our team. Repeated blocked requests pause AI features on an account for 24 hours.

Your photos and designs

Your uploads and generated images are stored privately in your account and are only shown to you; we don’t publish user creations or use them to train AI models. Please upload only photos you have the right to use. If you use results to market a property, label them as virtually staged or AI-edited where your local rules require it.

Report a problem

Email support@roommakerai.org with the image or link and a short description, or open a ticket with the Contact support button. We review every report; see the Acceptable Use Policy for response times and appeals.