How ReadSearch handles your information

What we keep, why we keep it, who helps us run the service, and how to delete all of it. No advertising, no tracking, and we never sell your data.

Who is responsible for your information

ReadSearch (readsearch.ai) is a research digest: every morning it picks the new papers that match your field and explains each one in a few short slides. ReadSearch is run by ReadSearch AI, Inc., a Delaware corporation (Delaware file number 10754535), which is responsible for your information under this policy (its "controller"). In this policy "ReadSearch", "we" and "us" mean ReadSearch AI, Inc. Our registered office is at 16192 Coastal Highway, Lewes, Delaware 19958, USA, and you can reach us about anything here at jad@eye.digital.

What we collect

Your account. Your name and email address. If you choose a password, we store it only as a salted hash and never see it again. If you sign in with Google or Apple instead, we receive your name, your email address, the account identifier that provider gives us, and from Google your profile picture. We never see your Google or Apple password.

What you tell us about your work. Your field, interests, techniques, role, university, research focus and short bio, your timezone, and your display choices (theme, colours, reading goal, streaks, listening voice). All of these are optional and you can change them in Settings.

How you use ReadSearch. How many questions you ask Ask ReadSearch and how many literature reviews you draft each day (only the counts, never the questions or reviews themselves; your recent chats and reviews are kept in your own browser, not by us, unless you save a review to one of your projects, and you can delete them there; an answer or review you choose to share is kept, with its question and sources, at its link until you stop sharing it), the papers you save to your library and any notes you write on them (including references you import from a file or from Zotero, whose key is used for that import only and never kept), the lab groups you create or join, with the papers you add to them and the comments you write there (which that group’s members see, with your name), the projects you create, with the papers, reviews, drafts and notes you keep in them (which members of a group you share a project with can read and edit), the systematic review screening projects you run, with their records, criteria and each screener’s decisions (which the project’s other screeners see once they have made their own decision), your ORCID iD if you add it to Your research, with the publications and citation counts OpenAlex lists for it and the authors you choose to follow, if you use the ReadSearch app and allow notifications, your phone’s notification address (removed when you sign out of the app or delete your account), the ratings you give, the papers you open and on which day, your reading progress (how far you got through each paper, so you can pick up where you left off on any of your devices), your recent searches (the last 20 search words you typed, which you can clear on Home at any time), the digests we showed you along with the profile details used to choose them, your research threads and the papers or reference lists you add to them, and any job posting you publish.

ORCID, if you connect it. Your ORCID iD and an access token, which we store encrypted and use only to read the works listed on your ORCID record. If you ask Settings to suggest interests from your publications, we look up the topics of those works in OpenAlex and show you matching interests; we keep nothing from it unless you add an interest yourself.

Your plan and payments. If you subscribe to Pro or Researcher on the website, you pay on Stripe's own page, and ReadSearch never sees or stores your full card number. From Stripe we keep your plan, your subscription's status (for example on trial, active, cancelling or payment due), whether you pay monthly or yearly, its trial, renewal and end dates, your Stripe customer and subscription identifiers, when you used your trial, and, if a partner discount applies, which partner's domain it came from. We also keep a log of the billing messages Stripe sends us (each message's identifier, its type and what we did with it), so that no payment is handled twice. If you buy a plan in the ReadSearch app, when it offers it, we keep the same kind of subscription record from the App Store or Google Play, never your payment details.

Technical information. To slow down password guessing we count sign-in attempts per IP address. Those counts live only in memory, for minutes, and are never written to our database. When something breaks, an error report is created (see Sentry below).

How we use it

  • To choose your daily papers and tailor their summaries and audio briefings to your field.
  • To send the codes that confirm your email address or reset your password, and, only if you switch it on in Settings, a weekly email of new papers in your interests. One click in any weekly email switches it off. We send no marketing.
  • To run your subscription: to start your trial, keep your plan in step with your payments, apply the right limits for your plan, lock or reopen your account when your plan ends or starts again, show your plan and its dates in Settings, check your email's domain against our partner list to apply a partner discount, and make sure no one is charged twice or given a second trial.
  • To keep accounts secure and the service working, and to fix errors.

We do not show advertising, we use no analytics or tracking tools, and we never sell or rent your information.

Your own documents, and patient information

In Ask ReadSearch you can attach a PDF, and in the citation finder you can paste a draft. This is your own work, and we handle it more strictly than the rest:

  • Our server reads the PDF in memory and sends its text straight back to your browser, where it stays while you work on it. We do not save the file, its text, your pasted draft or your questions about them to our disk, our database or our logs, and we keep no copy after we have answered.
  • To answer, each question sends the passages of your document nearest your question, together with the question, to OpenAI or Anthropic, through their own services and nowhere else. For a long document, the text is also sent in pieces to OpenAI so we can find those passages, and the numbers that stand for those pieces are kept in our server's memory only, marked to expire after 30 minutes, and never written to disk or to our database. If our AI connection ever points anywhere other than those two providers, these features stop and nothing is sent.
  • With OpenAI we ask it not to store the request. Both providers say in their terms for developers that they do not train their models on what is sent through their services. They may keep a request for a short time under their own rules, for example to detect abuse, and we cannot delete it from there.

Do not upload information that identifies a patient. ReadSearch is a reading tool for research. It is not built to hold health records, so please remove names, dates of birth, record numbers and anything else that could identify a person before you attach or paste a document. The same goes for any other private information about someone else.

Services that help us run ReadSearch

A few providers process information on our behalf, each only for the job described here:

  • Hetzner Online hosts our servers and database in Helsinki, Finland. Everything we store lives there.
  • Cloudflare carries all traffic to readsearch.ai and protects it from attacks.
  • OpenAI writes the paper summaries and audio briefings. For a summary it receives the paper's title and abstract, plus your field and the kind of work you do, so the summary fits you. To judge which papers match your interests it also receives your field, research focus, topics and the text of your research threads. When you use Ask ReadSearch, it receives your question, your last two questions and answers in that conversation, the papers found for it, and your role, field and research focus from your profile, so it can write the answer and pitch it to your work. When you search for a paper to summarise, cite or compare, it receives those search words to find it. A citation is made by ReadSearch itself and is sent to no one. It never receives your name or email address.
  • Anthropic reads a search you type when the search looks like it names an author, a title or a year, so we can understand what you mean.
  • Our own mail server delivers our emails. Resend is a backup for sign-in codes only: if it is used, it receives your email address and the code in the message.
  • Sentry collects error reports. We remove IP addresses, cookies, sign-in details and passwords before a report leaves our server, but a report can include what you had entered on the page that failed.
  • Stripe takes payments for Pro and Researcher on the website. On its payment page and in its customer portal it receives your email address, your card details and anything else you enter there, such as a billing address, and we tell it which ReadSearch account the subscription belongs to. Stripe acts partly on our behalf and partly as an independent controller, for example to prevent fraud and meet its own legal duties, and it handles your information under its own privacy policy at stripe.com/privacy. Stripe may email you receipts and notices about your subscription.
  • Apple and Google, only if you buy a plan in the ReadSearch app, when it offers it. The App Store or Google Play takes the payment under its own privacy policy (www.apple.com/legal/privacy/ and policies.google.com/privacy), and tells us only about the subscription and its dates.
  • Google or Apple, only if you choose to sign in with them, and ORCID, only if you connect it.
  • Research databases: the words you search for (in Search, and in Ask ReadSearch’s paper picker) are sent to OpenAlex and arXiv to find papers, without your name or email address. Your ORCID iD, if you add it to Your research or ask for interest suggestions, is sent to OpenAlex to find your publications.

Some of these providers are based in the United States, so your information may be processed there. We choose providers that protect it under their own published privacy and security terms.

Cookies and your browser

We use only the cookies the service needs to work:

  • Your sign-in: an httpOnly cookie that keeps you signed in for 30 days.
  • Your colour palette, remembered for a year.
  • Short-lived cookies that protect a Google or Apple sign-in while it is in progress, lasting ten minutes, or up to 14 days for Google's.

Your browser also keeps a few things on your own device, such as your theme, your voice, a half-finished onboarding, your listening queue and your recent Ask ReadSearch chats and reviews, which stay only in your browser. It also keeps a copy of your reading progress and recent searches so they work offline; the copy kept with your account is described above. You can clear what your browser keeps at any time in your browser's settings. We use no advertising or analytics cookies.

Stripe's payment page and customer portal are Stripe's own websites, and use Stripe's own cookies there.

How long we keep it, and deleting it

We keep your information for as long as you have an account. Sign-up and reset codes expire after 10 to 15 minutes, and a sign-in lasts 30 days from when you last used it.

You can delete your account at any time in Settings, under Account. That permanently removes your account and everything linked to it from our database, including your library, ratings, reading history, reading progress, recent searches, threads, ORCID connection and job postings, and it deletes your audio briefings from our server. Our encrypted backups are replaced on a 14-day cycle, so a deleted account disappears from them within 14 days.

We keep your plan and subscription records for as long as you have an account. When you delete it, we first cancel a subscription bought on the website at Stripe, so nothing more is charged, and delete your customer record there, then remove your plan and subscription records with the rest of your account. If Stripe can't be reached at that moment, we cancel it by hand as soon as we know, and you are not charged again. Records of the payments themselves, such as invoices and receipts, stay with Stripe for as long as the law requires them to be kept, for example for tax and accounting. The log of billing messages holds no name or email address and is kept only so that no payment is handled twice. A subscription bought in the App Store or Google Play is not cancelled by deleting your account, so cancel it in the store.

Information already passed to a provider, such as an error report, is kept under that provider's own retention rules.

Your choices and rights

You can see and correct most of your information yourself in Settings and on your profile, disconnect ORCID on the Threads page, and delete your account in Settings. For a copy of your information, or for anything you cannot do yourself, email jad@eye.digital and we will answer within 30 days. Depending on where you live, you may also have the right to object to or restrict how we use your information, and to complain to your data protection authority.

Where you live

Everyone gets the same tools: Settings to see and correct your information, Delete account to remove it, and email for anything else. The law where you live may add to that.

In the European Economic Area and the United Kingdom (the GDPR and the UK GDPR). We may use your information only with a legal ground. For most of what we do, it is that we need it to give you the service you asked for (a contract): your account, your daily papers, your plan and its payments, the codes we email you. For keeping ReadSearch secure and working, fixing errors and stopping abuse or a double charge, it is our legitimate interests, which we weigh against your rights. Where we ask, it is your consent: the weekly email, notifications in the app, and connecting ORCID. You can withdraw consent at any time by switching it off, and that does not undo what we did before. We keep some records because the law requires it, such as payment records at Stripe. You have the right to see your information, correct it, delete it, restrict or object to how we use it, and receive a copy you can move elsewhere, and to complain to your data protection authority, such as the Information Commissioner's Office in the UK. Some of our providers are in the United States (see above). Where the law requires a safeguard for that transfer, we rely on the ones those providers offer, and you can ask us which.

In California (the CCPA, as amended by the CPRA). We do not sell your personal information, and we do not share it for advertising across other companies' sites, because we show no advertising and use no tracking. You have the right to know what we hold and why, to have it deleted, and to have it corrected. We will not treat you differently for using these rights. You can do most of this yourself in Settings, and we will deal with the rest within 45 days. We confirm it is you by replying to the email address on your account, and you can ask someone else to act for you if they can show your permission.

In Canada (PIPEDA). We collect and use your information for the reasons in this policy, and by creating an account you agree to that. We ask again for the weekly email and for notifications. You can see and correct your information, withdraw your consent (which may mean we can no longer give you the service) and ask us about how we handle it. Your information is processed in Finland and the United States. If you are not satisfied with our answer, you can complain to the Office of the Privacy Commissioner of Canada.

Contact us at jad@eye.digital about any of this. Say where you live and what you would like, and we will answer within 30 days.

Security

All traffic is encrypted. Passwords and sign-in tokens are stored only as hashes, ORCID tokens are encrypted, and access to our servers is limited to the people who run them. No system is perfectly secure, and if a breach ever affects you, we will tell you.

Children

ReadSearch is made for researchers, clinicians and students, and is not meant for anyone under 16. We do not knowingly keep information about children, and we will delete any we find.

Changes to this policy

If we change how we handle your information, we will update this page and the date at the top. For a significant change, we will tell you by email or in the app before it takes effect.