Do you train on my images?

No. We do not use your photos, prompts or results to train AI models, and we do not sell personal data.

What happens when you press Generate

We do not run the AI models ourselves. Your prompt, settings and input files go to the gateway that serves the model you picked:

• OpenRouter for most image and video models,
• fal for lip sync, speech, music, upscaling, background removal and some image models.

The gateway runs the model itself or passes the request to the model's developer or a hosting partner. The result comes back to us and is stored in your account.

When a gateway needs one of your uploaded files, it gets a temporary link to that file rather than a permanent one.

Which companies make the models

The models on Snap AI Studio come from Alibaba, Black Forest Labs, ByteDance, Clarity AI, fal, Google, hexgrad, Kuaishou, Microsoft, MiniMax, OpenAI, Recraft, Runway, Sourceful, Stability AI, Sync Labs, xAI and ZhengPeng7. Each model page names the company behind it, and the privacy policy keeps the current list.

We send your content to these providers only to produce your result. Their own policies govern how long they keep request data; the privacy policy links to them.

Your own API key

If you saved your own OpenRouter or fal key, jobs on that gateway run under your own account with that provider, and its terms apply to them.

Who else can see your content

• You. Uploads and results are private to your account.
• The public, only if you choose. Share to gallery shows the result, the model name and your prompt. Your name and email are never shown. Unshare removes it.
• Our administrators, to run the service, investigate problems and enforce the content policy.
• Safety records. When a prompt is blocked or a provider's safety filter rejects a job, we keep the prompt, the reason, the model and the time.

The full detail, including every service provider we use, is in the privacy policy.