Privacy Policy

AI Research Aggregator collects AI papers and model releases from public sources, ranks them, and presents them differently depending on whether you read as a student, a professional or a researcher. You can browse most of it without an account.

The service is operated by Abhishek Kumar Srivastava, who is the data controller for the personal data described in this policy. You can reach him at srivastavaabhishek1234@gmail.com.

Account. When you sign in with Google or GitHub we store your email address, display name and avatar URL as supplied by that provider, along with the OAuth tokens needed to keep you signed in and a session record per browser.

Profile. Anything you enter yourself: current job title, industry, years of experience, skills, target role, learning goals, research interests, and — if you use the LinkedIn import — a text summary of your LinkedIn profile. All of it is optional.

Activity. Your bookmarks and notes, your generated study plans and which phases you have completed, article reactions (helpful / not relevant / already knew, plus an optional reason and note), a record of which articles were shown to you and in what position, product events such as opening the chat or completing onboarding, and a derived per-topic preference score used to rank your feed.

Chat feedback. When you rate a chatbot reply, we store the full text of that exchange — your message and the assistant's reply — together with your rating and any comment, so the answer quality can be reviewed. Rating a reply is always your choice.

Technical. IP addresses are used for rate limiting and for the free-query allowance given to signed-out visitors. We do not run third-party advertising or analytics trackers.

Every time the feed renders, we record which articles were shown to you and at what position in the list. This is an implicit signal — you do not have to click anything for it to be collected — so it is worth being precise about what it is used for.

Those records exist for exactly one purpose: ranking your feed. Together with your explicit reactions (helpful / not relevant / already knew) they are rolled up into a per-topic preference score — one number per tag, per user — and that score adjusts the order of what you see next. Knowing a card was shown and ignored is what makes an ignored topic sink; without the position it is recorded at, an article near the bottom of a long page would be indistinguishable from one you actively rejected.

It is not used for advertising, not used to build a profile for anyone else, not sold, and not shared with any third party. It does not feed the chatbot and is never sent to the model provider. We do not record reading habits for signed-out visitors at all — without an account there is nothing durable to attach them to, so they are simply not stored.

The underlying records are deleted after 90 days. The derived preference score survives, because it is a running number rather than a history of what you read. Deleting your account removes both.

Some features send data to third-party processors. Each is listed below with what actually reaches it.

We do not sell personal data, and we do not share it with anyone beyond the processors above.

Records of which articles were shown to you, and general product events, are deleted after 90 days by a scheduled retention sweep. The derived topic-preference scores that drive your feed survive that sweep — they are a running score, not a history of what you did.

Article reactions, bookmarks, study plans, your profile and your chat feedback are kept for as long as your account exists, because each is something you explicitly saved or submitted. Deleting your account removes all of it.

You can erase your account and everything attached to it at any time. Deletion is immediate and cannot be undone — there is no grace period and no backup copy you can ask us to restore from.

To delete your account: open Settings from the header menu, go to the Danger zone, and type DELETE to confirm. The typed confirmation exists so that a mis-click cannot erase an account. Deletion is automated and takes effect immediately — there is no ticket to raise, no email to send, and no one who has to approve it.

This is how we satisfy the right to erasure (the “right to be forgotten”) under Article 17 of the UK and EU GDPR, and equivalent rights under other privacy regimes. Because it runs without human review, it is not subject to the one-month response window such requests normally allow.

This removes:

• your user record, and with it your OAuth connections and sessions;
• your profile and LinkedIn summary;
• bookmarks and notes, and your study plans;
• article reactions, impressions and topic-preference scores;
• product events, chat feedback and any general feedback you submitted.

Content that was never linked to your account — anonymous feedback submitted while signed out — has no identifier tying it to you and is not affected.

Depending on where you live you may have the right to access, correct, export or erase your personal data, and to object to certain processing. Erasure is automated and available immediately from Settings, as described above; you can correct or clear most of your profile from the same place. For access, export or objection requests, email srivastavaabhishek1234@gmail.com.

If you object to your reading habits being used to rank your feed, deleting your account removes both the records and the derived scores. There is currently no way to keep an account while opting out of that ranking — the feed has no other personalisation mechanism to fall back on.

For any question about this policy, or to exercise a right that is not automated in the app, contact:

Please allow a few days for a reply — this is a solo-operated free beta, not a staffed support desk. Account deletion does not require contacting anyone; it is automated and immediate from Settings.

If this policy changes materially we will update this page and the date shown above. Because it ships with the application, its history is visible in the project's source repository. Your use of the service is also governed by the Terms of Service.