AI Output, References and Similarity Policy

Last updated: 18 October 2026

Lumy uses artificial intelligence to turn plain words into games, apps and websites. This policy explains how that works, what AI output can and cannot be trusted for, how references are used, and what we expect of you. It forms part of our Terms of Service.

1. How Lumy uses AI

Lumy uses a mix of leading AI systems, chosen and tuned by us, combined with our own game kit, prompts, quality standard, design reviews and tools. We choose which systems handle each step and may change them at any time without notice.
Builds run in Auto mode by default, which picks a model for each step. You may instead choose any one model for every step, on any plan. From Anthropic: Sonnet 5.5, Opus 5.5, Fable 5.1 and Haiku 4.5, and the earlier Opus 5, Sonnet 5, Opus 4.8 and Sonnet 4.6. From OpenAI: GPT-5.6 Terra, GPT-6.1 Sol and GPT 6 Astra, and the earlier GPT-6 Sol, GPT-5.5 and GPT-5.4. The models we use for images are our choice and are not listed.
AI is used to ask intake questions, write code and text, paint art, review designs, score quality (the Wow Score), suggest names and taglines, run Autopilot improvements, answer help questions, and power AI features inside the apps you build.

2. AI output has limits

Output may be wrong. It can contain bugs, security weaknesses, broken features, inaccurate facts, inappropriate material, or content that looks like existing work.
Output is not unique. Different people giving similar prompts can receive similar or identical results.
Output is not advice. Nothing Lumy produces is legal, medical, financial, safety or other professional advice.
Scores are opinions. The Wow Score and design reviews are automated judgements against our standard. They are not a promise of quality, safety, legality or success.
You must review. Test and check every project, and fix what is wrong, before you rely on it, publish it, sell it or give it to anyone.

3. Original work

Lumy is designed to make original work. It takes inspiration for feel, genre, mechanics and layout, and it will not, on its own initiative, reproduce another product's name, characters, logos, art, music, text or other protected expression. We may filter, rename or change output to reduce the risk of imitation.

4. References

You may give Lumy references: an uploaded image or screenshot, a link (for example a store, video or game page), or the name of a known product with an @-tag. Video references are not available yet.

References are for inspiration only: to show a style, genre, feel or layout.
Only share what you may share. Do not upload material you do not have the right to use, or private or confidential material.
We may refuse or limit references, strip names or elements from them, or decline requests that ask Lumy to copy, clone or imitate a specific product, brand, character or person.

5. Your requests are recorded

Every prompt, reference, @-tag, answer and instruction you give is stored with your project's version history, together with what changed. That record shows what you asked Lumy to make. We may use it to review reports, notices, disputes and appeals, and we may disclose it where the law requires or to defend claims.

6. You are responsible for similarity

If a project resembles an existing product, brand, character, person or work, that resemblance results from your requests and choices, and you are solely responsible for it. Before you publish, sell or distribute a project, you must make sure you have every right and permission it needs. You agree to cover any claims that arise, as set out in Section 21 of our Terms of Service. We may take down any project that we believe imitates or infringes someone else's rights, under our Copyright, Trademark and Takedown Policy.

7. Prohibited AI uses

In addition to our Acceptable Use Policy, you may not use Lumy or AI features in your apps to:

create deepfakes or synthetic media of real people without their consent, or present AI content as real footage or real events;
impersonate real people, brands or organisations, or pass off output as someone else's work;
produce content designed to mislead people about elections, health, safety or public events;
make fully automated decisions that have legal or similarly significant effects on people (for example about jobs, credit, housing, insurance, education or access to services) without human review and the safeguards the law requires;
build systems for social scoring, manipulation of vulnerable people, emotion recognition in workplaces or schools, predictive policing, or other practices that are prohibited or regulated as high risk, unless you comply with every law that applies;
extract our prompts, system instructions or kit through prompt injection or other adversarial inputs, or use output to build or train a competing builder or model.

8. If your app uses AI

If your project includes AI features for its users:

tell users clearly when they are talking to AI or seeing AI-generated content, wherever the law or fairness requires;
label AI-generated images, audio or video that could be mistaken for real;
do not present AI answers as professional advice, and add suitable warnings for health, legal, financial or safety topics;
put reasonable limits and moderation in place, and follow the Acceptable Use Policy in what your app lets users generate;
remember that AI calls inside your app are billed to your tokens, never to your users, and are limited to 300 a day per app.

9. Your content is not used to train our providers' models

Our AI providers are contractually barred from training their models on your content. We may use usage signals (such as versions you keep, ratings, errors and timings) and de-identified or aggregated content to improve Lumy's own quality systems, prompts and game kit, as described in our Terms of Service and Privacy Policy.

10. Human review

Automated safety systems may block, pause or flag requests, references and projects. People on our team review reports, notices, appeals and flagged content where needed, and may view the relevant parts of a project and its request history to do so. Staff access is limited and logged.

11. Changes

AI is changing quickly, and so will this policy and the systems behind Lumy. We will update this page when our practices change, as described in our Terms of Service.