Privacy Policy

1. Who this covers

This policy describes how Feedpikhandles personal data in the Feedpik web app, the Feedpik API, and the Feedpik Telegram bot. It does not cover the websites whose RSS feeds you subscribe to — when you open an article on a publisher’s own site, that site’s privacy policy applies, not this one.

Wherever this page says “we”, it means the operator of Feedpik. If you want the short version of everything below, it is this: we collect what the product needs to rank your feed, we run no analytics of any kind, and we do not sell or rent data to anyone.

2. What we collect

Account data

Your email address, your display name, and your avatar image if you set one. If you sign in with a password we store a hash of it, never the password itself. If you sign in with Google or GitHub we store the tokens that provider issues us instead.

Each active session record also holds the IP address and browser user-agent string it was created from. This is how “sign out everywhere” and suspicious-session detection can work at all; it is not used to profile you or to infer your location for any other purpose.

Things you create

• Your subscribed RSS feeds, folders, and personas — including the persona name, emoji, and description you write.
• Your ranking policies: the weights, filters, and instructions that decide how your feed is ordered, and every past version of them.
• Interest tags and blocked topics, whether you typed them or accepted a suggestion.
• Memory items, saved searches, competitor watchlists, briefings, and drafts you write in the draft workspace.
• Your conversations with the in-app assistant, stored as threads so you can come back to them.

Reading behaviour

When you read, save, like, skip, or give negative feedback on an article, we record that action against that article for the persona you are using. This is the raw material the ranking learns from — without it, the algorithm you own has nothing to learn from.

What we deliberately do not record: how long you spent on an article, how far you scrolled, where your cursor went, or whether a story was on screen. There is no dwell-time signal in Feedpik, because the ranking is not trying to maximise the time you spend here.

Newsletter inbox

If you use a Feedpik inbound address to receive newsletters, we store the messages sent to it — subject, body, sender name and sender address — the same way we store any other feed item, and mark those feeds private to your account.

Billing

Card details never reach our servers. Our payment providers handle the transaction and return us a subscription record: which plan, which period, the provider’s customer and subscription identifiers, and the email address and name you gave them at checkout. We store that record so we know what your account is entitled to.

Telegram

Only if you link the bot: your Telegram chat ID, numeric user ID, and username, plus your digest preferences. Unlinking clears them.

3. Tracking, advertising, and measurement

We do measure our own product

We would be flying blind otherwise, so we count things: pages viewed, accounts created, feeds added, which locked feature someone ran into, and whether a checkout completed. This runs on our own servers and writes to our own database. Nothing about it is sent to a third party, and there is no vendor with a copy.

To count visitors without following them, we do what a cookie would normally do — and then throw the key away. For each request we compute a one-way hash of a rotating daily secret, your IP address, and your browser’s user-agent string. The hash is what gets stored; your IP address is not stored at all. Because the secret changes every day, today’s hash cannot be matched to yesterday’s, so the record cannot become a history of you. Once you are signed in we use your account id instead, and stop computing the hash.

What this measurement never contains: your email address or name, which articles you read, what you searched for, or anything from inside the reader beyond the fact that a feature was used. It is a product-decision instrument, not a profile.

We also do not:

• Sell, rent, or trade personal data. There is no data-broker relationship of any kind.
• Take money to promote a story, a source, or a publisher. Nothing in your ranking is bought — subscriptions are the whole business model.
• Train models on your reading behaviour, your content, or your assistant conversations.
• Read your feed to build an advertising profile, because we do not sell advertising.

4. How the ranking uses your data

Feedpik scores each candidate article against your policy and your interests. Part of that comparison uses an interest vector derived from the articles you have responded to positively.

That vector is computed at query time, inside the database, and thrown away when the query finishes. There is no stored profile embedding sitting in a table waiting to be leaked, sold, or subpoenaed — the inputs are your own action rows, which you can inspect and delete.

Every score carries the reasons that produced it. You can open any ranked story and see which weights fired and by how much, and you can change them. The scoring core is published under the MIT licence as @feedpik/ranking — the same module the server imports, not a simplified illustration of it.

5. AI features and what they send

Summaries, translations, explanations, article Q&A, digests and drafts are produced by a third-party large-language-model provider. We do not run our own models.

These run on request, not in the background. When one runs, what leaves our servers is:

• For summarise, translate, explain and ask: the title and text of the article in question, plus your question if you asked one.
• For the assistant when it is tuning your ranking: your current interest tags and blocked topics, and the messages in that conversation.
• For Telegram digests: the titles and opening text of the articles in that digest.
• For ranking: article text is sent to an embedding model when an article is first ingested. This is about the article, not about you.

Your email address, account identifiers, and reading history are not included in these requests. If you are on a plan that uses your own API key, the request goes to your provider under your account and terms instead of ours.

6. What travels when you publish a Persona

A published Persona contains the recipe: the RSS feeds in it, and the tuning parameters — weights, filters, prompts. That is the whole payload.

Some sharing options let you include a derived summary — for example, which sources you engage with most — but only if you explicitly turn that on, and the summary is shown to you before it is published.

7. Services that touch your data

These are the third parties involved in running Feedpik, and what reaches each of them.

Each is used for the function described and nothing else. We do not authorise any of them to use your data for their own purposes, and we do not send them data that the function does not need.

8. How long we keep things

Your account data stays for as long as the account does, because a reading history that expired would quietly degrade the ranking you built. Two things do age out on a schedule:

• Articles— fetched items older than a year are deleted, along with everything derived from them. Anything you saved or liked is exempt, and so is each feed’s most recent item, so a quiet source never goes blank.
• Product-measurement records — deleted after 400 days. For anonymous visitors they were never linked to a person to begin with, and after the day they were recorded they cannot be, because the hash that grouped them is no longer reproducible.

Deleting your account erases the rest immediately: feeds, personas, ranking policies and their history, reading data, memory, drafts, briefings, newsletters, and assistant conversations. It is irreversible, so export anything you want to keep first — and if you have an active subscription, cancel it before deleting, since erasing the account here does not tell the payment provider to stop billing you.

9. Your rights, and the buttons that implement them

Depending on where you live you may have rights to access, correct, export, restrict, or erase your personal data, and to object to certain processing. We honour these requests regardless of where you live — drawing the line by jurisdiction seemed like the wrong instinct for a product whose entire argument is that your data is yours.

Several of these are already self-service:

• Export your feeds — standard OPML, importable by any other reader.
• Export your profile — your interests, memory items, and raw reading-action rows, as JSON.
• Export your ranking policies — JSON, including past versions.
• Delete your reading behaviour — clears the action rows the ranking learns from, from the app.
• Reset your profile — clears interests and memory items.
• Delete your account — erases everything, immediately, without asking us. It is at the bottom of the same settings screen as the exports above, which is the order we think you should meet them in.

For anything not on that list, email hello@feedpik.com. We aim to respond within 30 days. If you think we have mishandled your data, you may also complain to your local data-protection authority.

10. Security

Traffic is encrypted in transit. Passwords are stored only as hashes. API keys are stored only as hashes, shown once at creation, scoped to the permissions you choose, and revocable at any time. Sessions expire, and you can end them.

No service can promise it will never be breached, and we would rather say that plainly than imply otherwise. If a breach affects your data we will tell you and the relevant authority within the timeframes the law requires.

11. Children

Feedpik is not directed at children and is not intended for anyone under 16. We do not knowingly collect data from children. If you believe a child has created an account, write to us and we will remove it.

12. Changes to this policy

When the product changes what it collects, this page changes with it, and the date at the top moves. For changes that materially affect how we use data you have already given us, we will tell you by email rather than relying on you to re-read this page. Related: our cookie policy and terms of service.