AI training policies by provider
Each third-party AI provider that Artifio routes to has its own stance on whether your prompts and outputs may be used to train their models. We commit in Privacy Policy §5.4 to keep this page current within 30 days of any change we become aware of. When in doubt, consult the provider's own policy (link in the table) — their page is authoritative.
Current stance
OpenAI API (Sora, GPT, DALL-E)
Default not trainedAPI inputs are not used for training by default.
Provider privacy policy →Anthropic API (Claude)
Default not trainedAPI inputs are not used for training by default.
Provider privacy policy →Google (Veo, Imagen) via Vertex / Gemini API
Default not trainedVertex API inputs are not used for training by default.
Provider privacy policy →Black Forest Labs (FLUX)
Default not trainedAPI tier explicitly excludes training.
Provider privacy policy →Chinese providers (Kling, MiniMax, Alibaba/Wan, ByteDance/Seedance, Seedream, HiDream, Qwen)
Assume retained / may be usedTerms vary and change without notice. Users forwarding prompts to these providers should assume the provider may retain and use inputs under Chinese law. You can filter models by provider region in your account settings.
What Artifio does
Artifio does not train its own models on your prompts, uploads, or generated outputs. See Privacy Policy §5.3 and §5.4 for the full controller/processor characterisation.
Where a provider offers an opt-out flag on their API, Artifio sets it by default on the routed request. Providers whose default is to train are listed above with stance Assume retained / may be used.