Privacy at brainsto.ai

What we keep

• Your dilemma, the options you list, your answers to the facilitator’s questions and your view. This is what the models work on.
• What the models write back, and a technical record of each model call: which model, how many tokens, how long it took and what it cost. The technical record does not include your text.
• If you sign in: your lemed.ai account id, name and email, and an encrypted access token that lets us use your credits.
• Outside the ambassador program pages, the site uses no advertising cookies and does not track you. We use Plausible to count site visits without cookies or identifying you. To protect the site from bots, we use Cloudflare Turnstile. It checks your browser when you submit a dilemma, without advertising cookies or tracking you.
• To see which channels work, we count every visit to our site and its source (a link, campaign or referring site) on our servers. The count is anonymous: no IP address, no identifier and no third party. If you arrive through one of these links, our cookie stores only the source for 30 days. If you sign up during that period, the source is saved with your account.
• The Meta Pixel runs only on the ambassador program pages (the program page and the application form). It stores a cookie in your browser and reports page visits and application submissions to Meta. This lets us measure the program’s advertising on Facebook and Instagram. Meta processes this information under its privacy policy. The details you enter in the form are not sent to Meta, and the Pixel does not run on any other page on the site. You can block it in your browser settings or in your Meta account’s ad settings.

Which AI providers receive your text

• To run a discussion, your dilemma and your answers are sent to Claude (Anthropic), GPT (OpenAI), Gemini (Google), and Grok (xAI). Each provider processes the text under its own API terms and privacy policy.
• A short safety check of the text is also sent to GPT (OpenAI).
• Do not include names, ID numbers, addresses, medical details or anything else that identifies you or other people. You can describe the situation without them.

How long we keep it

• Discussions of signed-in users are kept until you delete them.
• A draft you wrote before signing in is deleted after 24 hours, or as soon as it becomes a discussion.

If you use your own API keys

• Your keys are stored only in your browser (localStorage). They are sent with each request that runs the models, used in memory for that request, and never written to our database, logs or error reports.
• The discussion then runs on your provider accounts, under your agreement with each provider. Remove the keys at any time on the keys page, or revoke them at the provider.

If you share a decision

• Sharing is off until you turn it on. A shared link displays only the decision card and the title you wrote, plus a short summary if you choose to include one. Anyone with the link can view it. Your dilemma, your answers and the discussion are never shared.
• The suggested title is written by a model that is told to remove identifying details; check it before you share. Shared pages are not indexed by search engines, and you can stop sharing at any time.
• If you choose to ask friends what they would do, the link also shows the options you wrote and the vote count for each option. Voting is anonymous, with no sign-in needed. To limit repeat voting, we store a hash of a random identifier from a browser cookie and a hash of the IP address, not the address itself. Votes are deleted when you stop sharing or delete the discussion.

If you publish on the public discussions page

• Public posting is separate from link sharing. It is off until you request publication and approve a specific draft (see the Terms of Use). Only a title and short summary written by an AI model without identifying details, along with the decision card, are published. Your dilemma, your answers, the discussion and your name are not published.
• To prepare the draft, your dilemma is sent to GPT (OpenAI). Before publication, the draft itself (not your dilemma) is checked by GPT (OpenAI) and Gemini (Google), then read and approved by a member of the brainsto team.
• We keep a record of every proposed draft, your consent and when you gave it, the review results and the decision. We also keep user reports, with a hash of the IP address rather than the address itself. These records are kept for as long as the discussion exists and deleted when you delete the discussion or all your data.
• You can remove your post at any time. Removal from our pages is immediate, but copies saved by others or stored in search engine caches are outside our control.
• Your published excerpt will only appear in brainsto’s social media posts if you give separate consent. Consent is optional. We record when you give consent and when you withdraw it. Once you withdraw consent, we’ll stop using the excerpt and delete our posts where possible. Other people’s shares and cached copies are beyond our control.

Deleting your data

• In My discussions you can delete a single discussion, or all your data at once. Deletion is immediate and permanent: the discussion, your answers, what the models wrote and the technical records of its model calls.
• Text already sent to an AI provider is subject to that provider’s retention rules.

Account and payments

• brainsto.ai is a lemed.ai service. Sign-in, your account and credit purchases are handled by lemed.ai, under the lemed.ai privacy policy. brainsto never sees your payment details.
• If you claim an offer through a sponsor’s link (for example, an insurance agent’s), the sponsor can see your first name, last initial, part of your email address and the date you claimed it. They can’t see your discussions, anything you write in them or how you use the service.

Sensitive topics

If your text suggests that someone may be in danger, no discussion takes place. Support phone numbers are shown instead. The check is automated. No person reads the text as part of it.

Contact

Questions and requests about your data: support@brainsto.ai

Automated safety checks

• All text submitted for discussion, along with the title and summary before sharing, is checked automatically by OpenAI’s moderation service. These checks aim to prevent prohibited use and protect the service. Checks run through our account, even when you use your own API keys.
• If text is blocked, it is not discussed and you are not charged. When a check is blocked or not completed, a brief record is saved: the account ID (or a random draft ID), the time of the check, the stage at which it ran, the categories flagged and what happened. The record does not include the text itself. It is deleted after 90 days or when the account is deleted.

Who can see your discussions

• Discussions are not published by default. To run a discussion, the text is sent to the AI providers listed above.
• People do not routinely read private discussions. There are three limited exceptions: when you contact support and ask us to review a discussion, when we have a valid legal obligation, or when we need to address a security incident.
• Anyone with a share link can open it and pass it on. You can disable the link at any time.

Age and emergencies

• The service is for people aged 18 and over.
• brainsto is not an emergency or crisis support service. It cannot guarantee that it will detect or intervene in a dangerous situation.

Refer a friend

When you open a referral link, the referral code is stored in a browser cookie for 30 days, solely to credit the person who invited you. To prevent abuse of the referral program, we also store a one-way fingerprint of your IP address (not the address itself).