Home › Legal › Content Handling

Content handling.

Exactly what happens to the text you submit — which systems score it, which third-party AI provider sees it, what we keep, and every switch you have over that.

Effective July 30, 2026 Version 3.0

The promise

Five rules, no exceptions:
  1. We only ever take the text you submit — never your files, contacts, photos, location, or anything you didn't hand us.
  2. We never sell your text, and we never give it to advertisers.
  3. The only third parties that see it are the AI providers that produce your result: Anthropic (Claude), and OpenAI in some configurations. They process it to answer the request and do not train on it.
  4. You decide what we keep. "Store my scan history" controls whether scans are saved, and is on by default. "Exclude my content from training" keeps your text out of our detection-model training, and is also on by default. Both are in Settings.
  5. You can delete any saved scan, or your entire history, at any time.

Who processes your text

Two kinds of systems see the text you submit, and we think you should know which is which.

In the mobile apps, scanning text with the camera is done entirely on your device — the photo never leaves it. Only text you then choose to analyze is sent to us.

The mobile apps ask for your permission before the first request that carries your text, and you can withdraw it later in Settings → Privacy & data. The full sub-processor list is in the Privacy Policy.

Lifecycle of your text

From the moment you click "Detect" to the moment the result is on your screen, here's exactly what happens:

1

You paste

Text travels over TLS to our API. Files are read server-side and only the extracted text moves on.

2

We process

Tokenized and scored by our own classifiers. For the writing tools, the text goes to Anthropic (Claude) instead.

3

We score

Sentence-level scores and the headline result are computed from the model output.

4

We return

The result is sent back over TLS and rendered on your screen.

5

We save — or don't

The scan and its text are saved to your history so you can reopen it. Switch "Store my scan history" off and nothing is saved.

We never write your text into our application logs, and error reports never carry it.

Training data — what we use, what we don't

Our detection models are trained on a mix of AI-generated text (produced by running public models ourselves), human-written text, and adversarial examples we generate to harden the model.

Your text: we do not use it to train our AI-detection models. "Exclude my content from training" is on by default for every account, new and existing, so this is the setting you already have unless you deliberately switch it off.

Never: we don't give your text to any third party to train their models, and Anthropic does not train on what we send them.

Saved history

Scans are saved to your history by default, so a result is still there when you come back to it. When a scan is saved:

API content

API requests are handled the same way as requests from the app: scored by our own models, with the writing endpoints sending text to Anthropic. The same account settings apply to API traffic on your account.

If we receive a valid legal request for customer data (subpoena, court order), here's what we do:

Breach protocol

If we suffer a data breach affecting customer data:

Questions or concerns? privacy@textsight.ai — we reply within 72 hours.

Questions we get asked

Do you train AI models on my submitted text?

No. “Exclude my content from training” is on by default on every account, so your text is not used to train, fine-tune or improve our detection models unless you deliberately switch that off in Settings.

How long do you keep my text?

If you are signed out, it is not kept at all — we process it and discard it. If you are signed in, scans save to your private history by default so you can reopen them, and they stay until you delete them. There is no automatic expiry on any plan, on any tier. You can delete an individual scan or your entire history at any time from Settings → Privacy, and you can switch saving off altogether.

Who can see my submitted text?

You, and the systems that process your scan. Your text is not sold, not shared with other users and not given to advertisers. A small number of TextSight staff can open stored scans when investigating abuse or answering a support request.

Is my text encrypted?

Yes. Traffic to and from TextSight uses HTTPS, and stored scan content is encrypted at rest by our managed database provider, including backups.

Do you use third-party AI providers?

Yes, for the generative features. Rewriting and the writing tools are processed by third-party LLM providers — Anthropic (Claude), and OpenAI in some configurations. The AI detector itself runs on models we operate. Every processor that can receive your text is listed on our subprocessors page.

Can I delete my scan history?

Yes. Individual scans are deleted permanently from your dashboard, and the whole history can be wiped from Settings → Privacy. Deleting your account deactivates it and ends access immediately. If you also want the underlying records erased outright, say so in your request — that is your right under GDPR and we will confirm when it is done.

Is TextSight GDPR or FERPA compliant?

We honour GDPR data-subject rights — access, rectification, erasure and portability — and the equivalents under UK GDPR, CCPA and India’s DPDP Act. FERPA applies to educational institutions handling student records; if you are using TextSight in a classroom, contact us before relying on it for anything FERPA covers.

Do I need a data processing agreement?

If you are processing personal data of EU or UK residents on behalf of an employer or institution, you probably do. Contact us with your requirements and we will work through them with you.

What happens to my data if I cancel?

Cancelling returns your account to the free plan. Your scan history is not deleted on cancellation — it stays until you delete it. Billing records are kept for as long as tax law requires.

Further reading