Bria

FAQ

Is Bria's content legally safe to use commercially?

Yes. All Bria models are trained exclusively on licensed data from Getty Images, Alamy, and Envato. Every output is covered by full IP indemnification, meaning Bria contractually protects customers against copyright claims from generated content. Bria is also SOC 2 Type II, ISO 27001, GDPR, and EU AI Act compliant.

What is VGL and why does it matter?

VGL (Visual Generative Language) is the structured language Bria's models are natively trained to read and execute. Instead of prompting with natural language, users direct every visual parameter explicitly: lighting, camera angle, composition, background, and objects, ensuring the same direction produces the same result every time.

Where does my data go?

Bria runs on Bria Cloud, your cloud (BYOC on AWS or Azure), on-premise, or fully air-gapped. Your inputs, outputs, and training data never leave your environment unless you choose a cloud option. Full data sovereignty is a deployment option, not a promise.

How does Bria fit into my existing workflow?

Bria integrates via API, MCP, Skills, SDK, or iFrame. For creative teams, native plugins exist for Photoshop, Figma, ComfyUI, Nuke, and Houdini. For developers and agents, every capability is independently callable and composable.

Can I train Bria on my own brand assets?

Yes, through Tailored Generation. You train a custom model on your visual IP, not shared data. The result is a brand-locked model that produces on-brand outputs by default, improving over time through feedback. You receive full model source code and weights.