Last Updated: August 18, 2026
NativeTavern is a local-first AI character chat application. NativeTavern does not operate an AI proxy and does not collect or store your conversations on NativeTavern servers. When you choose a remote AI feature, the app sends the data required for that feature directly from your device to the provider or endpoint you configured.
This policy explains what stays on your device, what may leave your device at your request, who may receive it, and how to control that sharing.
NativeTavern stores the following data locally:
- Characters, personas, system instructions, lorebooks, world information, memories, Data Bank content, and tool configuration
- Conversations, messages, bookmarks, and chat history
- Images, backgrounds, sprites, Live2D/Spine assets, audio, and generated media
- App preferences, provider configuration, API credentials, and backups
- Local capability and external-call diagnostics, which record destinations, data categories, status, and timing but not request content or API keys
This local data remains on your device unless you initiate an export, cloud backup, URL import, or remote AI request.
Before NativeTavern permits remote AI requests, the app displays a disclosure that identifies the data categories and supported recipients. You may choose either Allow remote AI or Use local AI only.
Your choice is saved on your device. You can withdraw or restore permission at any time under Settings > Remote AI data sharing. When permission is off, NativeTavern blocks remote AI requests before they leave the device. Local and private-network AI endpoints remain available. A materially changed disclosure will require a new choice.
The exact data depends on the feature you use:
- Chat and text generation: your current message, relevant chat history, character and persona content, system instructions, enabled lorebook/world information, memory or Data Bank excerpts, and enabled tool definitions or tool inputs
- Embeddings and semantic search: document text or excerpts selected for embedding
- Image generation: prompts, generation settings, and any source or reference image you provide
- Speech to text: recorded audio and selected language/model settings
- Text to speech: text to synthesize and selected voice/model settings; generated audio is returned to your device
- Authentication and request metadata: your API credential, model name, request settings, IP address, and standard network metadata required to make and authenticate the connection
NativeTavern sends this data only when you invoke or enable the corresponding remote feature. NativeTavern does not use it for advertising or sell it.
Depending on the provider and endpoint you configure, data may be sent directly to:
- Anthropic
- An OAI Compatible endpoint you configure
- OpenRouter
- Google Gemini
- DeepSeek
- Alibaba Cloud Qwen (DashScope)
- SiliconFlow
- Moonshot AI / Kimi
- Z.AI
- MiniMax
- Cohere
- ElevenLabs
- Microsoft Azure Speech
- Volcengine
- NovelAI
- Pollinations
- A self-hosted or custom endpoint whose address you enter
The active provider and endpoint are shown in the relevant AI configuration screen. If you enter a custom endpoint, the operator of that endpoint is the recipient and is responsible for its privacy and security practices.
Remote providers process data under their own terms and privacy policies. The supported providers publish privacy and security commitments intended to protect data they receive. NativeTavern connects directly to the provider using your account or API credential and does not control provider-side retention, training, or deletion. Review the selected provider's privacy policy and account data controls before use. Do not use a provider or custom endpoint whose protections do not meet your requirements.
Cloud backup is optional and user initiated. If you choose Google Drive or an available Apple/iCloud storage location, the selected backup file and the account information needed for authentication are handled by Google or Apple under their respective privacy policies. NativeTavern does not receive a copy of your backup through its own servers. You can sign out, delete cloud backups, or use local export instead.
When you request an import from a URL or open an external link, the destination website receives the requested URL and normal network metadata such as your IP address and user agent. Imported character data is stored locally after it is downloaded. The destination website's privacy policy applies to that request.
- Local data remains until you delete it in the app or uninstall the app.
- A local export remains wherever you save or share it.
- A cloud backup remains until you delete it from the selected cloud provider.
- Remote AI providers retain or delete transmitted data according to their policies and your provider account settings.
- NativeTavern cannot retrieve or delete data held by a provider because the request is made directly with your provider account or custom endpoint.
NativeTavern relies on device security for locally stored data and keeps API credentials on the device except when sending them directly to authenticate a request. Use HTTPS for remote custom endpoints, protect device access, and do not share exported backups or API credentials with untrusted parties.
NativeTavern may request:
- Camera or photo access: to select or capture character and background images, including location metadata that may be embedded in a selected photo
- Microphone and speech recognition: for speech features you invoke
- Files or storage: to import, export, and back up your data
- Network access: for remote AI, cloud backup, URL import, external links, and provider authentication
Permissions are requested when needed and can be changed in system settings.
You can:
- View, edit, export, and delete locally stored content
- Disable remote AI data sharing while continuing to use local endpoints
- Change or remove provider credentials and custom endpoints
- Delete cloud backups or disconnect a cloud account
- Revoke operating-system permissions
Depending on your jurisdiction, you may also have legal rights concerning data held by a remote provider. Submit those requests directly to that provider.
NativeTavern is not directed to children under 13 and does not knowingly collect their personal data. Users must meet the minimum age required by their jurisdiction and by each remote provider they choose.
We may update this policy when features or data practices change. We will update the date above and require a new in-app choice before remote AI sharing when a change materially affects the disclosure.
Questions or privacy requests may be sent to:
- Email: support@nativetavern.com
- Discord: https://discord.com/invite/URQvW2FvZa