Choose how you sign in, where your data lives, and which AI models you use, with transparent privacy options and user-owned memory.
Anuma is a private AI assistant designed to give you control over your data, your memory, and your AI experience. Choose how you sign in, where your information is stored, and which AI models power your conversations. Every setting is transparent, with choices between privacy, convenience, and capability.
Unlike traditional AI platforms that lock you into a single model and a single privacy policy, Anuma lets you access leading AI models including ChatGPT, Claude, Gemini, Grok, Qwen, Kimi, and more, while keeping ownership of your personal context and memory.
Privacy isn't all or nothing. You decide how you sign in, where your data lives, which AI models you use, and who else can see you, with the honest tradeoff behind every option laid out in plain language.
No email needed, no server backup, no closed-source models.
Sync across devices and use closed-source models like GPT, Claude. Your data is encrypted in transit and at rest, and those providers follow their own terms.
Privacy in Anuma isn't all-or-nothing. It's four dials you set yourself, each a choice between more private and more open. Try it below and watch your level update.
Your account identity. How much do we even know about you?
The tradeoff: Wallet sign-in needs no email or phone number, but you safeguard your own key. Email is easier to recover but links your account to a real address.
Whether an encrypted backup of your chats is kept on Anuma servers.
The tradeoff: Turning backups off erases the server copy, and you will not be able to access Anuma from other devices. Encrypted backups are encrypted in transit and at rest, but not end-to-end encrypted, so Anuma and its service providers may access them for operation, safety and legal compliance.
The engines behind your chats, and what they're allowed to see and keep.
The tradeoff: Open-source models are run by third-party providers whose own terms govern retention and training. Closed-source models are the most capable but may retain prompts per their own policies. Either way, Anuma does not use your conversations to train its own models by default.
Whether you share your location to appear on the map and meet people nearby.
The tradeoff: With this off, you can still browse the map without appearing on it, and your memory is only ever used for you. Turn it on to share your location and appear on the map, so people nearby can find you. It needs encrypted backup, and you can switch it back off at any time.
Nothing is locked in. Every dial can be changed from settings whenever you want. Privacy that grows and relaxes with you.
Your AI memory is the context that makes Anuma useful, and it belongs to you at every privacy level.
Not end-to-end, so Anuma and its providers may access it to run the service.
Your context follows you across every model you use.
View, edit, scope, or permanently delete any entry, anytime.
Toggle private for open-source only, or pick what each model sees.
Detected personal data is replaced with placeholders before the message leaves your device, then restored locally in the reply.
Email me at john.doe@example.com, my cell is 555-123-4567.
Email me at [EMAIL_1], my cell is [PHONE_1].
Turn on PII Redaction in Privacy settings and Anuma scans your chat messages on your device before they are sent. Detected personal details (emails, phone numbers, SSNs, credit cards, IP addresses, API keys) are swapped for placeholders, so the AI model only ever sees [EMAIL_1], never your real address.
When the response comes back, Anuma restores the real values locally, so your chat reads exactly as normal. The map between real values and placeholders stays in memory on your device and is never transmitted. No extra server, proxy, or vault in the path.
Most privacy tools redact at a server or gateway, where your data still reaches a third party first. Anuma does it on the device, so for the data it detects, the sensitive value is never sent at all. It is opt-in, off by default, and best-effort protection for common, structured personal data.
Privacy-first tools keep your data safe, but most make you choose between private models and powerful ones. Anuma gives you both, plus memory you actually own.
| Anuma | Proton Lumo | Brave Leo | DuckDuckGo | Venice.ai | |
|---|---|---|---|---|---|
| Sign-in without an email address | Yes | Yes | Yes | Yes | Yes |
| No model training on your conversations by default | Yes | Yes | Yes | Yes | Yes |
| Open-source models | Yes | Yes | Yes | Yes | Yes |
| On-device PII redaction | Yes | No | No | No | No |
| Closed-source models (GPT, Claude & Gemini) | Yes | No | Limited | Limited | Limited |
| Memory you own and export | Yes | Limited | Limited | No | Limited |
| Memory that follows you across models | Yes | No | No | No | No |
| Choose your own privacy level | Yes | Limited | No | No | Limited |
Comparison based on each provider's publicly available documentation as of June 2026. Features and pricing change often, so please verify current details on each provider's own site.
No vague promises. Here is exactly where your data lives, how it is encrypted, and who may access it.
Your conversations, memories, and files are stored on the device you are using, in an on-device database (IndexedDB in the browser, SQLite on mobile), and each wallet gets its own isolated instance. Encrypted backups of your conversations are also held on Anuma's servers.
Media files live in a sandboxed filesystem that other pages cannot touch. You can turn backups off in Settings, which erases the server copy. Prompts you send are passed to third-party model providers, whose retention and training follow their own terms.
Your messages and conversations are encrypted in transit and at rest. They are not end-to-end encrypted unless a specific feature says so, so they may be accessible to Anuma and its service providers for operation, safety and legal compliance.
Prompts are sent to the third-party model provider you choose. Each provider’s retention and training are governed by its own terms, not Anuma’s, and Anuma does not control them.
Want your data to survive a lost laptop? Encrypted backups of your conversations are held on Anuma servers. They are encrypted in transit and at rest, but not end-to-end encrypted, so they may be accessible to Anuma and its service providers for operation, safety and legal compliance.
You can turn backups off in Settings, which erases the server copy. You can also delete individual or all conversations at any time.
Anuma does not use your conversations to train its own models by default, and you can turn backups off to erase the server copy.
Anuma does not build advertising profiles from your conversations.
Anuma does not use your conversations to train its own models by default. Model providers handle their own training under their own terms.
Open-source and closed-source models are run by third-party providers whose own terms govern retention. Anuma does not control them. For the least server-side copy, turn backups off.
Private AI is artificial intelligence you can use with control over your identity, conversations, and personal data, including how you sign in, where it is stored, and whether it is used for training. With Anuma, privacy is a spectrum you control: you choose how you sign in, whether backups are on, and which models you use. Your conversations are encrypted in transit and at rest, and you can turn backups off or delete your data at any time.
Anuma gives you real control over your privacy. Your conversations and memories are stored on your device, and encrypted backups are held on Anuma servers until you turn them off. Messages are encrypted in transit and at rest, but not end-to-end encrypted, so they may be accessible to Anuma and its service providers for operation, safety and legal compliance. Anuma does not use your conversations to train its own models by default, and open-source models are run by third-party providers whose own terms govern retention. You can dial privacy up or down.
The most private setup on Anuma is wallet login, backups off, and open-source models. In that mode you do not give Anuma an email address, no backup copy is kept on Anuma servers, and open-source models are run by third-party providers whose own terms govern retention and training. You can relax any of these dials for more convenience or capability, and Anuma shows you the exact tradeoff for each choice.
Yes. You can start using Anuma as a free private AI chat app, with your conversations encrypted in transit and at rest and backups you can turn off. Paid plans add more usage and access to additional AI models, but privacy works the same on every plan.
The best private AI chatbot is clear about where your data lives and who can access it, and gives you real controls. Anuma encrypts conversations in transit and at rest, and lets you turn off backups, delete conversations, or delete your account in the app.
You can keep your identity out of it, but it is not fully anonymous. You can sign in with a wallet rather than an email, so you do not have to give your name, email, or phone number to start a chat. Anuma still collects technical information such as your IP address and device details. Conversations are encrypted in transit and at rest, but not end-to-end encrypted.
Yes. Anuma is available as a private AI chat app on Android and iOS, as well as on the web. Your memory and conversation history sync across devices through encrypted backup, so you can start on your phone and continue on desktop.
Not necessarily. Turning backups off keeps the most data on your device, but it is not the only way to stay private. Anuma also offers encrypted backup, which lets you use your account across devices. Backups are encrypted in transit and at rest but not end-to-end encrypted, so they may be accessible to Anuma and its service providers for operation, safety and legal compliance. Private AI is about knowing who can access your data, not only about where it is stored.
Frontier models are the most capable but the least private, because your prompts are sent to their providers, whose retention and training are governed by their own terms. Anuma does not control them. For the least server-side copy, turn backups off. Open-source models are also run by third-party providers, so their own terms apply.
Yes. Nothing is locked in. Every dial (sign in, backup, and models) can be changed from settings whenever you want. Your memory works at every privacy level and follows you across models and devices, so you can tighten or relax privacy without losing your context.
Start fully private and relax the dials whenever you want, or never.