Trust

What happens to the content you check

People send us essays, evidence photos and private footage. This page says plainly what we store, for how long, and who processes it on our behalf. For legal terms, see our privacy policy.

No training on your data

Text, images and video you send are never used to train our models.

Files are not kept

Uploaded images and video are deleted after the check. We keep the result, not the file.

Zero retention on request

For institutional and enterprise agreements we switch off detection history entirely.

Contracts for regulated teams

A data processing agreement and dedicated capacity are available, or run our models on your own GPUs and servers.

What we store, by product

Website

Stored

Your detection history: the type of check, the verdict and score, and the first 100 characters of a text or the file name of an image or video. Switched off under zero retention.
PDF reports and shared result links you create, until you delete them.
Account and credit balance.

Not stored

Uploaded images and audio files.
Uploaded videos, which are held in temporary storage only until the check finishes.

Self-serve API & MCP

Stored

Usage counters for billing: number of calls and words per month. They contain no content.

Not stored

The text, images and video you send.

Enterprise API

Stored

One result record per file: score, verdict, model version, processing time and, for video, duration and frames analysed. No file content.

Not stored

The files themselves.
Anything beyond what your contract allows: retention is set per agreement.

Providers that process data for us

These companies process data only to run the service. Enterprise contracts can restrict this list further.

Provider	What for	Used for
Vercel	Hosts the website and API; temporary storage for video uploads until the check finishes	All
DigitalOcean	Runs our AI detection models for text, images and video	All
RunPod	Dedicated GPU capacity, reachable only through our own gateway	Enterprise
Google Firebase	Accounts, database for history and credits, site analytics	Website & self-serve
Paddle	Payments, invoicing and tax as merchant of record	Paid plans
OpenAI	Language models for thesis review and written analysis; extra image analyses; OpenAI’s provenance check on images and audio	Website & self-serve, not Enterprise

For enterprise customers

Detection runs on dedicated GPU capacity, reached only through our gateway.
No third-party language models: enterprise requests never go to OpenAI.
Retention, regions and data handling are set in your contract and data processing agreement.
Self-hosted option: run our models on your own GPUs and servers, on-premise or in your cloud. Files then never reach us.

Certifications

We do not hold SOC 2 or ISO 27001 certification yet. We are happy to complete your security questionnaire, sign a data processing agreement and walk your team through how data moves through the service.

Questions about data handling

Ask for our data processing agreement, a completed security questionnaire or a call with your security team.