Privacy

Last updated 26 September 2026

The short version

Your documents are never saved. Names, dates of birth, record numbers and addresses are removed on our server before anything is read. What we keep is a summary of the diagnosis and treatment, stored in a form that has no field for a name. We do not ask who you are unless you choose to create an account, and then we hold only an email address.

What happens to a document you upload

1. It is received by a server in Google Cloud and held in memory. It is never written to disk and never stored.
2. Identifying details are removed: names, dates of birth, medical record numbers, addresses, phone numbers, email addresses. Full dates are reduced to month and year.
3. The scrubbed text is sent to Anthropic to be read into a structured summary. That summary has fields for diagnosis, stage, biomarkers, genomics, treatments and lab values — and no field capable of holding a name.
4. The original text is discarded when the request ends.

The scrubbing is best effort. The structure is the guarantee: even where the scrubber misses something, there is nowhere in our database for it to be stored.

What we store

• The de-identified summary of your records.
• Which trials matched, and the reasoning for each.
• An email address, only if you create an account. It is kept in a separate part of the database that is never joined to your medical information.
• Payment records are held by Stripe. We never see your card details.

What we never store

• Your documents.
• Your name, date of birth, address, phone number or medical record number.
• Anything clinical in our operational logs. Logs record counts, timings and costs — never a diagnosis, and never the name of a file you uploaded.

Email

We never put medical information in email. Not a diagnosis, not a trial name, not a count of matches — a count implies a diagnosis to anyone who sees a notification on a lock screen. Our messages say only that something has happened and that you can sign in to see it.

Companies that process data for us

• Google Cloud — receives uploaded documents, under a signed Business Associate Agreement.
• Anthropic and OpenAI — read the de-identified text. Neither receives identifying information.
• Supabase — stores the de-identified records.
• Vercel — serves the website. Uploaded documents do not pass through it.
• Stripe — payments. Resend — email.

How long we keep things

Searches are kept for 180 days from the last time you use them. If you come back, the clock restarts. If you do not, they are deleted along with every verdict and question derived from them.

Deleting your data

Email privacy@trialsfinder.ai and we will delete everything associated with your account. Because we do not hold your name, we can only act on a request from the email address on the account, or from someone holding the recovery code.

Whether HIPAA applies

Probably not, and we would rather say so than imply otherwise. HIPAA governs healthcare providers, health plans and the companies working on their behalf. When you upload your own records to a service you chose, that relationship generally falls outside it. We hold a Business Associate Agreement with Google Cloud anyway, and we designed the system so that the question matters as little as possible.

If a clinician uses this on a patient's behalf, the relationship is different and HIPAA may well apply. That route does not involve an email address.

Children

This service is for adults. Trials for children are included in the corpus, but the account must belong to an adult.

Questions: privacy@trialsfinder.ai · Terms