Trust & data

Privacy policy

In effect 14 September 2026. This page describes the system as implemented: what is stored, who can reach it, who else processes it, and how it is removed.

At a glance

Your studies are visible only to the account that created them. That rule is enforced inside the database, not merely in the interface.
Uploads sit in private storage: no public links, no directory listing, no anonymous reads.
Card details never reach DebateLens. The payment processor holds them; we keep only a paid flag.
The site sets no advertising or cross-site tracking cookies and reports to no analytics vendor.
Deleting a study removes its records; deleting your account removes everything else, on written request.

Who is responsible for the data

DebateLens is operated by Dr. Hussain Al Sharoufi, the researcher whose framework the platform implements and who is credited on the Citation & provenance page. It is not operated through a company, and there is no data-protection officer or support queue: requests about your data are handled directly by the operator.

The operator has an owner-only administration panel. Through it the operator can see each account's email, name, institution, sign-up date, subscription status, sign-in times (with the browser used), activity counts, and the full content of studies, transcripts and reports; and can create, suspend, change the tier of, or remove accounts. This access is used to run the service, give support, prevent misuse and carry out your requests, and for no other purpose.

What we store about you

An account consists of the email address you register with, a display name if you choose to give one, and an authentication record. Signing in with Google brings over the email address and profile name Google returns. Passwords are stored only as hashes.

Once you register, the account also carries a paid timestamp and a reference to the payment session that produced it. That reference exists so the site can tell that study creation has been unlocked; it contains no payment instrument data.

What your studies contain

Everything you place in a study stays in that study: uploaded recordings and audio, transcripts and the turns they are divided into, the coded observations across the five analytical domains with their confidence and review states, gaze and repetition measurements, factual-claim records and their sourced evidence, reliability samples, reports, and drafts of the article engine.

Each of those records is bound to the account that created it. Another signed-in researcher cannot list, open or search them.

What is never held here

Card numbers, expiry dates, security codes or bank credentials — the payment processor holds those.
Readable passwords.
Identity documents or personal financial details; the site has no reason to ask for them.
Any material from an account other than your own, at any level of access.

Automated processing, and your choice in it

An automatic analysis sends extracts of your material to the models that transcribe, code and evaluate it. Their output is stored as a draft, labelled machine-derived, and excluded from every computed figure until you accept or qualify it.

If you would rather nothing of yours leave the installation for automated processing, upload a transcript instead of a recording, or code the study manually — both routes are supported and produce the same records.

How long material is kept

There is no expiry and no scheduled purge: material remains in your account for as long as the study that holds it exists. Deletion is immediate once you carry it out.

One honest caveat: the hosting layer may retain encrypted backups for a limited period. A record removed from the live system stops being reachable straight away, while copies in backups age out on the host's own schedule.

Deleting your material

Deleting a study removes its turns, observations, evaluations, claims, evidence, reliability records and reports.
Deleting a media asset removes the stored file itself, not only its entry in the study.
Exporting first is always open to you: CSV and JSON exports carry the counted evidence and coded records out of the system.
To have an entire account and everything attached to it removed, write to the address below. The request is carried out by hand: the account record, its uploads, its studies and every derived record go together.

Cookies and local storage

The browser keeps a sign-in session so you stay signed in between visits, and remembers the theme you chose. Those are functional, not behavioural. The site sets no advertising cookies, places no cross-site trackers, and sends no usage data to analytics vendors.

Your requests

You can ask what is held about you, ask for a correction, ask for deletion, or take your data out yourself through the exports in the app. None of these require an explanation, and none of them are routed through an automated system.

Broadcast material and public speech

Televised debates are public political speech, and analysing them is ordinary scholarly work. That does not transfer broadcast rights: you remain responsible for holding the right to upload and retain any recording you place in the system, and for the terms on which you publish extracts of it.

What this statement does not claim

No system that holds research material can honestly be described as risk-free, and this page makes no claim of formal security certification. What it does state is the access model as implemented, the parties that touch your material, and the way to get that material removed.

Changes to this policy

The version in effect is shown at the top and in the footer of every page (14 September 2026). A material change to what is stored, who processes it, or how it is deleted is stated here before it takes effect.

How researcher details and study data are protected

The controls below are the ones actually in place in this installation, each enforced where the data lives rather than in the screen you are looking at.

Row-level security on every table: Each of the nineteen tables that hold research data — projects, speakers, turns, observations, evaluations, claims, evidence, reports, exports and reliability records — has row-level security enabled. Every read and write is matched against the account that owns the row before the database returns anything.
Per-account upload storage: Recordings, audio and transcripts are uploaded to a private bucket whose access rules require each file to sit inside a folder named for the account that uploaded it. A request for another account's folder is rejected by the storage layer itself.
Server-side checks on every request: Analysis, coding and payment operations run through server functions that re-establish the signed-in account before touching any record. Hiding a control in the interface is never treated as the access check.
Encrypted in transit and at rest: Every request to the site travels over HTTPS. The managed database and object storage that hold your material encrypt stored data; passwords are kept only as cryptographic hashes, never in readable form.
No anonymous path to research data: No table grants read access to unauthenticated visitors, so no study, transcript or upload can be fetched without signing in as its owner.
Nothing is published automatically: Studies, reports and exports stay inside your account. Exports are generated only when you press the button, and nothing is posted anywhere without your action.

Every party that touches your material

There are three, and they are listed here in full. Your material is not distributed beyond them, and is never sold or shared for advertising.

Automated analysis models: Transcription, turn segmentation, framework coding and the comparative evaluation. They receive extracts of your material — audio segments, video frames, transcript text — solely to produce a result for the study you started. DebateLens does not use your material to train any model, and the providers' own terms govern how long they keep what they receive.
Stripe (payments): Runs the registration checkout and holds card details, receipts and refunds. DebateLens stores only a paid timestamp and a session reference; it never sees a card number, expiry date or security code.
Hosting and platform infrastructure: The database, storage and authentication services that run the installation, under their own processing terms.
Nobody else: No advertising networks, no analytics vendors, no data brokers, and no resale or sharing of researcher or study data.

Raising a concern

Write to support@debatelens.org. Requests about access, correction or deletion are answered in person.