AI Transparency
How AI is used, how outputs are labelled, and what you must disclose.
Which AI we use
Images and video: Black Forest Labs FLUX models by default. Texts, vision analysis and moderation: OpenAI models (e.g. GPT-6 Luna, GPT-6 Astra). Platform admins may route operations to other providers (e.g. Google Gemini); the active models are visible in each workspace’s job history.
Labelling of outputs (EU AI Act Art. 50(2))
Every generated or edited image keeps the provider’s Content Credentials (C2PA) where present; otherwise we embed IPTC metadata “DigitalSourceType: trainedAlgorithmicMedia”. Downloads are named with an “_AI” suffix. The app shows “AI-GENERATED” / “AI-EDITED” badges. Images pushed to Shopify receive an “AI-generated image” alt text.
Your disclosure duties (deployer)
If you publish content that shows realistic people, places or events that do not exist or did not happen as shown (e.g. synthetic models, edited photos of real people), you may have to disclose that it is AI-generated (EU AI Act Art. 50(4); New York synthetic-performer disclosure for ads; consumer-protection law). Do not present AI models as real people or AI images as exact representations of fit, colour or material.
Human oversight
Outputs are suggestions. Review every image, video and text before publishing — in particular product facts, claims and the fidelity of try-on results.
Safety measures
Prompt moderation, blocking of sexual content and content involving minors, minor detection on real-person photos, consent records for uploads of real people, automatic deletion of such uploads, no biometric templates.