Legal
Privacy Policy
This explains what Dianoetic collects, why, who else sees it, and what you can do about it. We've tried to keep it plain.
The short version
- Your messages, and any images or files you attach, are sent to AI providers (Groq, Hugging Face, Cloudflare, and OpenRouter, Mistral, Cerebras or NVIDIA when configured) to generate replies and images.
- If you have an account, we save your conversations so you can come back to them. Guests' chats stay in their own browser.
- Saved conversations may be used to improve Dianoetic's models. You can turn that off at any time.
- We don't sell your data, show ads, or use tracking cookies.
- Community messages are public. Everything else you write is private to your account.
- You can delete any conversation, or your whole account, whenever you like.
What we collect
If you create an account
- Your username and the date you signed up.
- Your password, if you set one, stored only as a salted, one-way hash (PBKDF2). We can't see or recover the password itself.
- Google or GitHub sign-in: if you sign in with Google, we receive your Google account ID, email address and whether it's verified, and we store the ID and email to recognise you next time. If you sign in with GitHub, we store your GitHub user ID; we use your GitHub username to suggest your Dianoetic username. We don't receive your Google or GitHub password, and signing in doesn't give us access to your email, files or repositories.
- Discord sign-in: if you sign in with Discord, we store your Discord user ID and, if Discord has verified it, your email address, and we use your Discord username to suggest your Dianoetic username. To check that you're in the Dianoetic Discord server, we ask Discord for the list of servers you're in, look for ours, and keep nothing else from that list. If you aren't a member, no account is created. We don't receive your Discord password or messages.
- GitHub connection: if you separately connect GitHub for the repository plugin, we store an access token encrypted at rest so Dianoetic can read your repositories and open pull requests when you ask. You can disconnect it in Settings at any time.
- Your conversations: your messages, the replies, which model and mode answered, how long it took, and any ratings you give. This includes images and text files you attach and images Dianoetic Images creates for you (images are saved as reduced-size copies).
- Your training setting: whether your conversations may be used to improve our models.
- Billing status: for paid plans, we store Stripe customer and subscription IDs, status, and your current plan. Stripe handles card details; Dianoetic does not store your card number.
- Community messages: what you post in the community, with your username and the time. These are public: anyone visiting the community page can read them.
If you use Dianoetic without an account
We don't store your conversations on our servers. They're kept in your browser's local storage so they're there when you come back, and they're gone if you clear your browser data.
Everyone
- Technical data: your IP address is used briefly to rate-limit requests and protect the service. Our host, Cloudflare, keeps standard request logs.
- Service statistics: we count replies, failures and images created per model per day. These counts contain no message content and aren't linked to you.
- Usage records: when you get a reply, we record the time and how many credits it used, so we can apply plan limits. For guests these are tied to a one-way hash of your IP address, not the address itself. Records older than a week are deleted.
- One cookie,
dn_session, which keeps you signed in for up to 30 days. It's essential to accounts working, so we don't ask for consent to it. There are no analytics or advertising cookies. Your appearance and default-model choices are kept in your browser's local storage.
Who else processes your data
Most replies come from open models run by the providers below. When Dianoetic's own fine-tuned model answers, it runs on Cloudflare. To answer you, each message and the conversation before it are sent to one of these providers:
- Groq, which runs several of our models.
- Hugging Face, which routes requests to inference partners such as Novita, Together AI, Fireworks, DeepInfra and others.
- Cloudflare, which hosts the site, stores account data and runs Workers AI models.
- OpenRouter, if configured, which routes some fallback AI requests.
- Cerebras, if configured, which answers some fallback AI requests.
- Mistral AI, if configured, which answers some fallback AI requests. On Mistral's free tier, Mistral may use those requests to improve its models.
- NVIDIA, if configured, which answers some fallback AI requests and checks incoming messages for attempts to override the assistant's safety instructions. NVIDIA logs requests to its API for security purposes.
- Stripe, if you purchase a plan, which processes your payment and manages your subscription. Stripe receives the information you enter at checkout.
If you use Google, GitHub or Discord to sign in, that company also knows you signed in to Dianoetic, under its own privacy policy. Image conversion in Dianoetic Images' tools happens entirely in your browser and isn't uploaded.
Voice input uses your browser's built-in speech recognition. In Chrome and Edge, your browser's maker (Google or Microsoft) turns your speech into text; Dianoetic only receives the text. Read aloud runs in your browser.
Each handles data under its own privacy terms. Which provider answers a particular message depends on availability. Please don't share things you wouldn't want processed by these services, such as passwords, financial details or health records.
How we use your data
- To generate replies and run the service.
- To save your conversations and show them across your devices.
- To keep Dianoetic secure and prevent abuse.
- To improve Dianoetic's models, if you allow it (below).
Improving our models
If your training setting is on, which is the default for new accounts, we may use your saved conversations and ratings to:
- evaluate which models answer best,
- build datasets to fine-tune Dianoetic's own models, and
- turn highly rated coding answers into reusable knowledge that helps Dianoetic answer similar questions. Email addresses, phone numbers and keys are removed automatically, and a member of the Dianoetic team reviews and rewrites each one into a general lesson before it's used. Nothing is added without that review.
You can turn this off at any time from the account menu in the sidebar. Once it's off, your conversations won't be included in any future dataset. Data already used in a dataset or a trained model can't practically be removed from it. Guest chats are never used, because we don't store them.
How long we keep it
Conversations stay until you delete them or delete your account. Each community channel keeps its most recent 1,000 messages; you can delete your own at any time. Deleting your account cancels an active subscription and removes your username, password hash, linked Google or GitHub sign-ins, GitHub and Microsoft connections, API keys, custom plugins, feedback, knowledge suggestions made from your answers that the team has not yet reviewed, sessions, conversations (including attached and created images), billing IDs and community messages from Dianoetic immediately. Stripe may retain payment records under its own policies. Copies may remain in our database provider's point-in-time backups for up to 30 days before they expire. If you claimed the free Thinker trial, we keep your Discord account ID so the same Discord account can't claim it again.
Your choices and rights
- Delete a conversation: hover over it in the sidebar and choose delete.
- Stop training use: switch it off in the account menu.
- Delete your account: in Settings, under Account, confirmed with your password (or your username, if you sign in with Google, GitHub or Discord).
- Get a copy of your data, or ask anything else: email us (below).
Depending on where you live, you may have further rights, such as to access, correct, or object to processing of your data. Contact us and we'll help.
Children
Dianoetic isn't intended for anyone under 13, and we don't knowingly collect their data. If you believe a child has created an account, contact us and we'll remove it.
Security
Passwords are hashed, sessions are stored only as hashes, all traffic is encrypted over HTTPS, and requests that change your account must come from Dianoetic itself. No system is perfectly secure, but we take reasonable steps to protect your data.
Changes
If we change this policy in a meaningful way, we'll update the date above and point it out in the app.
Contact
Questions or requests: privacy@dianoetic.tech