Privacy Policy
Last updated 21 October 2026
What Thesisroom collects, who receives it and for how long. Your name is redacted from your thesis before any analysis, and it is never used to train models.
On this page
- 1. Who we are
- 2. What we collect
- 3. Why we use it, and on what legal basis
- 4. How your thesis is processed
- 5. Emails, reports and links you share
- 6. Who receives your data
- 7. Where your data is processed
- 8. Abuse prevention and the free-run record
- 9. How long we keep your data
- 10. Cookies and browser storage
- 11. Your rights
- 12. Age
- 13. Security
- 14. Changes to this policy
- 15. Contact
This policy explains what Thesisroom collects, why, who receives it, how long it is kept, and the rights you have over it. Thesisroom is a research quality platform, and your thesis is a private academic document: the policy is built around protecting it.
Thesisroom is operated by Giolalytics, a sole proprietorship registered in France. For anything, including your data rights, write to support@thesisroom.com.
1. Who we are
Thesisroom is operated by Giolalytics, a sole proprietorship registered in France. Giolalytics decides how your personal data is used, which makes it the controller under data protection law.
There is one address for everything, data rights included: support@thesisroom.com.
2. What we collect
By where it comes from:
- Your account. Your name and email address. If you sign in with Google, the name, email address and profile picture Google gives us. The level, field and stage you add. The country your account was first opened from, read from your connection (we keep the country, not your IP address).
- Your documents. The thesis, chapters, proposals and reference lists you upload, the text taken from them, and a redacted copy of that text (section 4).
- What you type. Your Defense Practice answers, typed or dictated; passages you paste into Find Sources; what you write in Daily Desk, your journal and your feedback notes; your Research Compass searches and claims; an address you enter to send a report.
- Papers you give Paper Match, whether you upload them or choose an open-access paper.
- What you do in Thesisroom. The runs you start and their results, and our record of which outside services each request used (its size and time, never its content).
- Payments. Stripe handles them. We keep a customer reference, your plan, and the amount and date of your first payment. We never see or store your card number.
- Abuse-prevention signals. A device fingerprint your browser computes on its own, codes made from the content of your uploads, your country, and counts of sign-ups and email codes kept under codes made from an IP address or an email address.
- Consent records. Your cookie choice and its history, your signup agreement, and your choice about product updates.
- Attribution. The campaign, referral link, Trybe click id or Meta click id that brought you, if any.
- Signup context for Meta. Only when your consent allows advertising at the moment you sign up: the IP address and browser you signed up from and Meta's browser cookies, sent to Meta (section 6).
3. Why we use it, and on what legal basis
Each use has its own legal basis:
- To run the service you asked for: the rooms, your saved history, your account, your plan, and the emails that come with them (sign-in codes, receipts, billing and account notices, results you asked for). Basis: our contract with you.
- To keep the service secure and fair: rate limits, the signup checks, the free-run record (section 8), device and content fingerprints. And to count, ourselves, which pages and links lead to sign-ups. Basis: our legitimate interest in protecting the service and understanding how it is found.
- Analytics and advertising measurement (Google Analytics, Google Ads, the Meta Pixel and Conversions API, Trybe, TinyStats), and product-update emails. Where consent comes first (the European Economic Area, the United Kingdom, Switzerland, Brazil, Quebec, and anywhere we cannot tell), the basis is your consent, which you can withdraw at any time. Elsewhere, it is our legitimate interest, and each one stops when you turn it off.
- To improve the product using anonymised, aggregated patterns, only if you turn on Help improve Thesisroom in Settings. Basis: your consent, optional and revocable at any time from your settings. Nothing is being collected under it today.
- To keep billing records. Basis: our legal obligations.
4. How your thesis is processed
When a room runs, depending on the room, the whole redacted text or excerpts of your thesis are sent to one of the AI providers in section 6. Before that, a redaction step removes:
- your full name and your surname, as they appear on your account, wherever they appear;
- email addresses and phone numbers;
- runs of six to ten digits, which is how a student ID usually looks;
- the title page, the declaration and the acknowledgements, when it can recognise them;
- header and footer lines that repeat your name.
Your supervisor's name, your institution's name and other people's names can remain where they appear outside the title page, for example in the body of a chapter. Authors you cite stay, because they are part of the research.
Not redacted: what you type into the rooms (your Defense Practice answers, a passage you paste into Find Sources, what you write in Daily Desk, a Research Compass search or claim), and papers you give Paper Match. They are sent as you wrote them.
The original text of your uploads is kept so that you can read it in your account and so that a redacted copy can be made from it. It is never sent to an AI provider: a thesis that has no redacted copy yet gets one before any room reads it.
Your thesis text is never used to train any AI model. We do not train models on it, and each provider says so in its own terms: Anthropic's commercial terms say it may not train models on customer content; Google Cloud's service terms say it will not use customer data to train or fine-tune any model without permission; Tinfoil's privacy policy says it does not use content to train models.
What each provider keeps: Anthropic deletes inputs and outputs within 30 days; Google Cloud may hold a prompt in a cache for up to 24 hours and may log prompts to detect abuse; Tinfoil keeps nothing after it answers.
Tinfoil: The text of your uploads and what you type, when a room runs on it, and a code made from your account id that lets it reuse its own cache. Before any text is sent, Thesisroom checks a cryptographic proof that Tinfoil's sealed environment is genuine and runs the approved model; if the check fails, nothing is sent. Tinfoil says it keeps no content after it answers and does not train on it.
Academic databases receive the text of your references, their identifiers, and search terms built from your claims and your searches, so that references can be checked and papers found. They never receive your name or your email address.
6. Who receives your data
We do not sell your personal information. These providers process it for us, on our instructions and only for the purposes here:
| Provider | What it receives | Why | Where, and how transfers are covered | Its privacy page |
|---|---|---|---|---|
| Vercel | Every request to the site: your IP address, browser, the page and your cookies. Payment notices from Stripe pass through it on their way to our database. | Hosting the website and its server functions. | United States (server functions run in Washington, D.C.); pages are served from the nearest of its locations worldwide. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | vercel. |
| Convex | Everything your account holds: your profile, your uploads, their text and redacted text, every room's results, and our logs. | Our database, file storage and backend. | Convex, United States. Transfers: Standard contractual clauses in its data processing agreement. | convex. |
| Google sign-in | When you choose Continue with Google: your name, email address and profile picture from Google. | Signing you in. | Google LLC, United States. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | policies. |
| Cloudflare Turnstile | When you sign up with an email and password: your IP address and browser details, to tell a person from a bot. | Protecting sign-up from automated abuse. | Cloudflare, Inc., United States. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | cloudflare. |
| Stripe | Your email address, your account reference and the plan you buy. You enter your card on Stripe's own page; we never see the card number. | Payments, invoices and the billing portal. | Stripe, United States. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | stripe. |
| Resend | The address an email goes to and the email itself (and, for a supervisor report you send, the report). | Sending email. | Resend, United States. Transfers: Standard contractual clauses in its data processing agreement. | resend. |
| Anthropic (Claude) | Redacted thesis text, and what you type into the rooms. No account details are sent with it. Anthropic's commercial terms say it may not train models on customer content, and it deletes inputs and outputs within 30 days. | The analysis behind the rooms. | Anthropic, United States. Transfers: Standard contractual clauses in its data processing agreement. | anthropic. |
| Google Cloud Vertex AI (Gemini) | The same kind of text as Anthropic, when a run is routed to it. Google's service terms say it will not use customer data to train or fine-tune any model without permission. | The analysis behind the rooms, as an alternative provider. | Google LLC; processed in the United States. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | cloud. |
| Tinfoil | The text of your uploads and what you type, when a room runs on it, and a code made from your account id that lets it reuse its own cache. Before any text is sent, Thesisroom checks a cryptographic proof that Tinfoil's sealed environment is genuine and runs the approved model; if the check fails, nothing is sent. Tinfoil says it keeps no content after it answers and does not train on it. | Analysis inside sealed, verifiable hardware (confidential computing). | Tinfoil, Inc., United States, and its hosting providers. Transfers: Standard contractual clauses in its data processing agreement. | tinfoil. |
| Academic databases: Crossref, OpenAlex, Semantic Scholar, Unpaywall, PubMed, DataCite, OpenAIRE, HAL and DOAJ | The text of your references, their DOIs and titles, and search terms built from your claims and your searches (Semantic Scholar receives a claim you type in Research Compass). Never your name or email. | Checking that references exist, finding papers and open versions, and mapping citations. | Each database's own country, in the United States and Europe. Transfers: None of your identity is sent; these are public research services. | crossref. |
| Google Analytics and Google Ads | With analytics or advertising consent: the public pages you visit, your browser and device, and cookies, from your browser; a sign-up and a first payment (amount and a reference), from our server only, because the signed-in app loads no trackers. | Measuring how the site is used and whether our advertising works. | Google LLC, United States. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | policies. |
| Meta Pixel and Conversions API | With advertising consent: views of the public pages, from your browser; a sign-up and a first payment (amount and a reference), from our server only, because the signed-in app loads no trackers, with a one-way code made from your email address and the IP address and browser you signed up from. | Measuring whether our advertising leads to sign-ups and payments. | Meta Platforms, United States and Ireland. Transfers: The EU-U.S. Data Privacy Framework and Meta's data transfer terms. | facebook. |
| Trybe | With advertising consent: page loads (address, referrer, browser, screen size) and a random visitor id; on your first payment, if a Trybe creator's link brought you, the amount and a reference. Never your email or name. | Measuring which creators and campaigns lead to visits and payments. | Trybe Tech, Inc., United States. Transfers: Trybe's own terms; it names no transfer mechanism. | jointrybe. |
| TinyStats | With advertising consent: a script that, on a visit from one of its campaigns, records the visit with the page, the referrer and a short description of your browser (its type, language and time zone). | Measuring a campaign. | Not stated by the provider. Transfers: The provider publishes no privacy terms we could find. | None published |
| YouTube | Only after you press play on a video: the usual YouTube request data, in its privacy-enhanced mode. | Showing a video. | Google LLC, United States. Transfers: The EU-U.S. Data Privacy Framework (with its UK and Swiss extensions), and standard contractual clauses. | policies. |
| Unsplash | When you open a blog post with an Unsplash photo: your browser loads the photo from Unsplash, which receives the usual request data (your IP address, your browser and the page that asked for it). | Showing the photos on blog posts. | Unsplash Inc., Canada. Transfers: The provider's policy says it will meet the GDPR's transfer requirements where they apply, and names no mechanism. | unsplash. |
| Your browser's speech service | When you dictate an answer: your voice, handled by your browser's own speech recognition (in Chrome, that is Google's service). Thesisroom receives only the text. | Dictation in Defense Practice. | Your browser's vendor. Transfers: Your browser vendor's own terms. | None published |
A reference that links to a web page is checked by visiting that page, and an open-access paper you choose for Paper Match is read from where it is published. Those sites see a request from our server, not from you.
7. Where your data is processed
Most of our providers are in the United States, so your data leaves the European Economic Area, the United Kingdom and Switzerland. Each transfer is covered by the safeguard named for that provider in section 6: the EU-U.S. Data Privacy Framework, where the provider is certified, and standard contractual clauses in its data processing agreement.
Our website's server functions run in Washington, D.C., in the United States; its pages are served from the location nearest to you. Google Cloud's AI processing for Thesisroom runs in the United States.
8. Abuse prevention and the free-run record
When an account is deleted, free or paid, we keep a short record for 12 months so that deleting and signing up again cannot be used to get new free runs. It holds a one-way code made from the email address, codes for the device and the uploaded documents, and which free runs were already used; an account that ever paid is recorded with every free run used. The raw email address and the documents themselves are never stored in it, and nothing else about you is kept with it.
Each record is deleted automatically 12 months after the account was deleted. Records made from 16 October 2026 use an email code keyed with a secret only Thesisroom holds, so it cannot be turned back into your email address; records made before then use an unkeyed code and are deleted when their 12 months end. The document codes cannot be turned back into your documents.
You may ask us to delete your hashed identifiers. We will do so, with the consequence that the protection is lifted: a thesis you previously uploaded could then be re-uploaded by anyone matching the same fingerprints.
The device fingerprint is computed in your browser by an open-source library, with no outside call. Sign-up counts by IP address are kept under a code made from the address, for 7 days.
9. How long we keep your data
| What | How long |
|---|---|
| Your account, your uploads and everything the rooms made from them | While your account exists, or until you delete them. |
| Your account, after you ask to delete it | Deleted at the first nightly run after a 7-day grace period, so about 8 days after you ask. Signing in during the grace period cancels the deletion. |
| One thesis, when you delete it | Deleted at once with its stored files and every room's results on it. Notes in your journal move to another version of the same thesis, or go with the last one. |
| Your saved work after a paid plan ends | Readable for 30 days, then locked until you subscribe again. It is not deleted. |
| The free-run record a deleted account leaves (section on abuse prevention) | 12 months, for free and former paid accounts alike. |
| Referral, coupon and creator commission records | Kept without any link to your account, because they are another person's earnings or a one-time code's use. |
| Payment records and invoices | Kept by Stripe under its own legal duties; Stripe controls them. |
| Failed sign-in attempts (by a code made from the address) | 24 hours. |
| Upload attempts (rate limit) | 2 days. |
| Signup counters by a code made from your IP address | 7 days. |
| Email-code rate limits (by a code made from the address) | 30 days. |
| Our record of emails sent to an address with no account (a sign-up code, an invitation) | The address is kept 30 days, then replaced by a keyed code. |
| Our record of emails sent to your account, and of which providers each request used (never the content) | While your account exists. |
| Your consent choice for cookies | 13 months, then you are asked again; the history goes with your account. |
| Copies held by the providers below | Each provider's own retention, described with it. |
11. Your rights
You can:
- see and correct your profile at any time in Settings;
- delete a thesis, or your whole account from Settings;
- get a copy of your data: write to support@thesisroom.com and we answer within a month;
- object to a use based on our legitimate interest, or ask us to restrict a use;
- withdraw consent at any time: Cookie settings in the site footer for cookies, the Product updates switch in Settings or the unsubscribe link in any such email, and the Help improve Thesisroom switch in Settings;
- complain to your data protection authority; in France, that is the CNIL.
Declining optional consent never changes the product you receive.
In the United States: we do not sell personal information for money. Measuring our advertising with Google, Meta, Trybe and TinyStats may count as sharing under some state laws. Do not sell or share my information, in the site footer, turns it off, and so does a Global Privacy Control signal from your browser.
12. Age
Thesisroom is for people aged 16 or older, and we ask you to confirm it when you sign up. If you believe someone younger has an account, write to support@thesisroom.com and we will delete it.
13. Security
The site and its backend are reached only over encrypted connections (HTTPS). Your thesis is redacted before any AI provider sees it, and no request to an AI provider carries your account details. Every request to an outside service is logged with its size and time, never its content. Uploads are checked by their real file type and size before they are opened.
A session ends after 24 hours without use, and an open tab signs you out after 6 hours without activity. Repeated failed sign-ins are slowed and then stopped.
No system is perfectly secure. If something goes wrong that affects your data, we will tell you and the authority as the law requires.
14. Changes to this policy
When this policy changes in a way that matters for your data, we update the date at the top and, where appropriate, tell you. A new version of the Terms or this policy is asked of new accounts at sign-up.
15. Contact
Questions about this policy, or requests about your data: support@thesisroom.com. A real person answers.