Privacy Policy
This Privacy Policy explains what personal data we process when you use Probator.ai, why, on what legal basis, who we share it with, how long we keep it, and the rights you have under the EU General Data Protection Regulation (GDPR) and Portuguese law. It applies to the website, the editor, the API and our emails, including the emails we send to people at schools, universities, publishers and companies who are not users (see Business contacts in section 3).
1. Who is responsible
The controller of your personal data is Spawncore, Unipessoal Lda, NIPC/NIF 516797778, registered office at Largo do Esteiro 6, 2050-261 Azambuja, Portugal. Full company details are in our Legal Notice. For any privacy question or request, write to privacy@probator.ai.
When a school, company or other organisation uses Probator.ai to check documents that contain personal data of other people (for example student essays or candidates' CVs), that organisation is the controller of that data and we act as its processor under our Data Processing Agreement. In that case, please direct requests about that data to the organisation.
2. What data we process
| Category | Examples | Source |
|---|---|---|
| Account data | Email address, optional name, interface language, plan, date you accepted the Terms, sign-in times | You |
| Sign-in data | Hashed one-time codes; passkey public keys and device names; hashed session tokens; browser user-agent of each session | You, your device |
| Content | Texts and files you check or save, extracted text, reports and results, document titles. On plans with certificates, a PDF or Word file you check is kept for 24 hours so that a certified copy stamps your original file, then deleted | You |
| Usage data | Checks run, word counts, credits used, how many times you analysed each text (counted by its fingerprint, shown on certificates), API keys and connected-app tokens (stored as hashes), the names of apps you connected, and when they were used | Generated by the Service |
| Billing data | Stripe customer ID, plan, subscription status, billing period, tax country and VAT number; the IDs of the payment notifications Stripe sends us. Card details are handled by Stripe and never reach us. | You, Stripe |
| Purchase and consent records | For each purchase: what you bought, the amount, the date and the Stripe payment references; the time you confirmed, before paying, that you ask us to start at once and that you are 18 or older or buying for an organisation, with the version of that wording and of the Terms; when we sent the order confirmation; and any withdrawal and refund | You, Stripe, generated by the Service |
| AI agent registrations | When an AI agent registers to connect to your account: the email address entered to claim it, the name the agent gives itself, the IP address it registered from (for IPv6, only the first half of the address), an approximate location derived from that address (city and country), and the registration's status and times | The agent, you |
| Technical data | IP address, request logs, error reports. IP addresses are used transiently for security and rate limiting and are not stored, except for AI agent registrations (above) and the pseudonymous trial counter (below) | Your device |
| Trial data (no account) | A pseudonymous key: a salted hash of your IP address that changes every day, and the credits used that day | Generated by the Service |
| Certificates | Certificate ID, results, word count, language, text fingerprint (SHA-256), and the document title and your name if you choose to show them; the signed PDF; the certificate code; and your certified document, a copy of it (made when you certify a PDF or Word file you checked, or when you download it: a PDF of your file or text with the certificate ID and verification address at the top of each page, and the ID in its metadata), kept so you can download it again and deleted with your account. If you add a human review record: the reviewer name you give, your statement, the time, fingerprint and AI likelihood of your first check, and the share of words changed (the text of the first check is used to compute this and is not stored) | You, generated by the Service |
| Business contact data | For people we write to at schools, universities, publishers and companies: name, work email address, role and organisation as published on the organisation's website, the page where we found them, the emails we sent and any reply, including a request to stop | Your organisation's public website, you |
| Correspondence | Messages you send us and our replies. Emails sent to our addresses (such as hello@probator.ai and support@probator.ai) are stored, with any attachments, in our mailbox with Cloudflare (stored in the EU) | You |
We do not ask for special categories of data. Your content may nevertheless contain any kind of information; please avoid submitting sensitive personal data unless it is necessary.
3. Why we use it and our legal bases
| Purpose | Legal basis (GDPR art. 6) |
|---|---|
| Providing the Service: running checks, storing documents, managing your account, credits, API keys and connected apps, signing you in | Performance of our contract with you (art. 6(1)(b)) |
| Billing, invoicing and tax compliance | Contract (art. 6(1)(b)) and legal obligation (art. 6(1)(c)) |
| Proving the contract and complying with consumer law: recording what you confirmed at checkout, your purchases, order confirmations, withdrawals and refunds | Contract (art. 6(1)(b)) and legal obligation (art. 6(1)(c)) |
| Connecting AI agents to your account at your request, and preventing abuse of agent registration | Contract (art. 6(1)(b)) and our legitimate interest in a secure service (art. 6(1)(f)) |
| Security, abuse prevention, rate limiting, enforcing credit limits and our Terms; processing each payment notification from Stripe only once | Our legitimate interest in a secure and fair service (art. 6(1)(f)) |
| Trial without an account | Legitimate interest in letting people try the Service while preventing abuse (art. 6(1)(f)) |
| Issuing certificates and running the public verification page for them | Contract (art. 6(1)(b)): you ask us to issue a verifiable certificate |
| Essential service emails (sign-in codes, order confirmations, withdrawal acknowledgements, billing notices, changes to the Terms) | Contract (art. 6(1)(b)) and, for order confirmations and withdrawal acknowledgements, legal obligation (art. 6(1)(c)) |
| Answering your messages and handling complaints | Contract or legitimate interest (art. 6(1)(b), (f)) |
| Writing to people at organisations for which the Service may be useful (business contacts), and keeping a record so that we never write again to anyone who asked us to stop | Our legitimate interest in presenting the Service to organisations whose work it is relevant to (art. 6(1)(f)), within the electronic-marketing rules of the recipient's country |
| Defending legal claims | Legitimate interest (art. 6(1)(f)) |
Anonymous statistics. We keep daily totals of checks (for example, how many checks found hidden characters or which share of AI-detection results were inconclusive, by language) and publish them monthly on our blog. These totals contain no texts, fingerprints, account or device data, and groups of fewer than 20 checks are never published, so they are not personal data.
Business contacts. We sometimes write to schools, universities, publishers, translation companies and similar organisations, and to people there whose role relates to checking or teaching writing (for example the director of a writing centre or the managing editor of a journal). We only use an address that the organisation itself publishes on its website, either as a general contact or next to the person's name and role. We never buy lists or guess addresses, and we do not write to students or children. Each email says where we found the address and how to stop further emails, and we send a first email and at most one follow-up. If you reply "no" or ask us to stop, we will not write again: we delete your details and keep only your email address on a do-not-contact list so that we can honour your request. The data comes from your organisation's public website (GDPR art. 14).
We do not sell your personal data and do not use it for advertising. We do not send marketing emails to users of the Service; the only promotional emails we send are the business-contact emails described above. We do not use your content to train AI models, ours or anyone else's.
Where we rely on legitimate interests, you can object at any time (see section 8).
4. AI processing and automated decisions
Your text is analysed by automated systems, including our own detection model and third-party language models that run on Cloudflare Workers AI or are reached through OpenRouter, as described in our AI Policy. These results are information for you; we do not make decisions about you based on them, and the Service is not designed to make decisions that produce legal or similarly significant effects on individuals (GDPR art. 22). Customers who use results to make decisions about others must involve human review.
5. Who we share data with
We use carefully selected service providers ("sub-processors") that process data on our behalf under contracts that require confidentiality and security. The current list, with their role, location and transfer safeguards, is on our sub-processor page. In summary:
- Cloudflare hosts the Service, its database, file storage and our mailbox, sends our emails and receives the emails sent to our addresses. On Workers AI it runs the embedding models (bge-m3 and EmbeddingGemma) used by our detection model and the open model gpt-oss-20b for the rewrite test. Workers AI inference is not guaranteed to take place in the EU (see section 6).
- OpenRouter routes the text being checked to third-party language models: OpenAI GPT-6 Luna for the expert reading (served by OpenAI, Microsoft Azure or Amazon Bedrock) and Mistral Small 3.2 for grammar and spelling suggestions (served by DeepInfra, Parasail, Venice or Mistral AI). We require routing only to providers that do not collect the text under OpenRouter's provider data policies, so it is not used to train models.
- Stripe processes payments. Stripe is an independent controller for some data, for example for fraud prevention.
- TOConline, our certified invoicing software, receives the name, address, country, VAT number and amounts needed to issue each invoice, which we must keep under Portuguese tax law.
- Google (Google Workspace) hosts the mailbox from which we write to business contacts and in which we receive their replies.
- Anthropic (Claude) is used to research organisations' public websites, help draft emails to business contacts and keep our record of them.
- Search and academic services (Brave Search and Serper for the web, OpenAlex and Semantic Scholar for academic sources) receive short query phrases taken from your text during a plagiarism check, with no account data or other information that identifies you. They are third-party services that we query only to search for possible sources.
Certificates are public by design. If you create a certificate, anyone who has its ID or code (for example from the PDF, its QR code or the footer of your certified document) can see its details on our verification page: the results, word count, language, text fingerprint, issue date, any human review record and, if you chose to show them, the document title and your name. The analysed text is never shown on that page.
We may also disclose data where required by law or by a court order, to protect our rights, or to a company that acquires the Service (in which case this policy continues to apply).
6. International transfers
Our database (accounts, credits, purchases, certificates and document records), the documents you save and our mailbox are stored with Cloudflare in its European Union jurisdiction (D1 and R2): that data is stored, and the database runs, only in the EU. Requests are handled by Cloudflare's global network, so a request may be processed briefly in the data centre closest to you, inside or outside the EU. The AI models used for checks run on Cloudflare Workers AI, also in Cloudflare's network; Cloudflare does not guarantee that this inference takes place in the EU, so the text being checked may be processed briefly outside the EU. Text sent through OpenRouter for the expert reading or grammar suggestions is processed in the United States, or in France when Mistral AI serves the request. For Cloudflare, Stripe, OpenRouter and the model providers it routes to, we rely on an adequacy decision (including the EU–US Data Privacy Framework for certified companies) or on the European Commission's Standard Contractual Clauses, together with additional safeguards such as encryption in transit. The search services used for plagiarism checks (Brave Search, Serper, OpenAlex and Semantic Scholar) are in the United States and receive only short query phrases from your text, with no account data or other information that identifies you. Business contact data is processed by Google and Anthropic, including in the United States, under an adequacy decision (the EU–US Data Privacy Framework) or the Standard Contractual Clauses. You can ask us for a copy of the relevant safeguards.
7. How long we keep data
| Data | Retention |
|---|---|
| Account and usage data | While your account exists; deleted when you delete your account |
| Saved documents and their reports | Until you delete them or your account |
| Text checked without saving | Not stored in your account. A copy of the report may stay in an internal cache for up to 24 hours so that an identical check is not charged twice. Plagiarism results are never cached. |
| Files kept for certified copies (PDF or Word files you checked) | 24 hours, then deleted automatically |
| Comparison library fingerprints | Until you remove them (Account › Privacy & data) or delete your account |
| Certificates, their PDFs and certified copies | Until you delete your account. A revoked certificate stays visible as "revoked" until you delete your account or we delete revoked certificates; after that, the verification page shows the certificate as not found |
| One-time sign-in codes | 10 minutes (deleted once used) |
| Sessions | 30 days, or until you sign out |
| Trial counter (no account) | 1 day (the pseudonymous key changes daily) |
| AI agent registrations (claim email, IP address, approximate location) | Unclaimed: deleted 1 to 2 days after the 24-hour claim window ends. Declined or disconnected: deleted 30 days after that. Approved: until you disconnect the agent (then deleted 30 days later) or delete your account |
| IDs of Stripe payment notifications | 90 days (security and to process each notification only once) |
| Server logs | Up to 30 days |
| Database backups | Up to 30 days, after which deleted data is no longer recoverable |
| Billing and tax records | 10 years, as required by Portuguese tax law (held by Stripe and in our accounts) |
| Checkout consent and purchase records (including withdrawals and refunds) | 10 years after the purchase, with our billing records |
| Business contact data | Up to 2 years after our last exchange. If you ask us to stop, we delete it at once and keep only your email address on a do-not-contact list |
| Correspondence, including emails sent to our addresses | As long as needed to handle your request, then up to 3 years, unless you ask us to delete it earlier |
8. Your rights
You have the right to:
- access your personal data and receive a copy;
- rectify inaccurate data;
- erase your data ("right to be forgotten");
- restrict processing in certain cases;
- data portability: receive your data in a structured, machine-readable format;
- object to processing based on legitimate interests;
- object at any time to receiving business-contact emails (direct marketing), without giving a reason: reply "no" to any of our emails or write to privacy@probator.ai and we stop;
- not be subject to decisions based solely on automated processing with legal or similar effects.
You can exercise several of these rights yourself: Account → Privacy & data lets you download all your data as a JSON file and delete your account. For anything else, write to privacy@probator.ai. We answer within one month. We may ask you to confirm your identity, for example by replying from your account's email address.
You also have the right to lodge a complaint with a supervisory authority, in particular the Portuguese Comissão Nacional de Proteção de Dados (CNPD), www.cnpd.pt, or the authority in your country of residence.
9. Security
We protect your data with encryption in transit (TLS) and at rest, passwordless sign-in (email codes and passkeys), hashed storage of codes, session tokens, API keys and connected-app tokens, access controls, and per-account separation of stored documents. Only authorised staff can access production systems, and only when necessary. No system is perfectly secure; if a personal-data breach is likely to put you at risk, we will inform you and the supervisory authority as required by law.
10. Children
The Service is not intended for children under 13. You must be at least 16 to create an account; people aged 13 to 16 may only use the Service with the consent of a parent or guardian or through an institution that has obtained it, as described in our Terms. Paid plans and top-ups are only for people aged 18 or over, or for purchases on behalf of an organisation; this is confirmed at checkout.
11. Cookies
We only use cookies and similar storage that are strictly necessary for the Service, such as the session cookie that keeps you signed in, or that remember preferences you set, such as your language. We do not use analytics or advertising cookies. See our Cookie Policy.
12. Changes to this policy
We will publish any update on this page with a new date. If the changes are material, we will also tell you by email or in the Service before they take effect.