SafeHuman AIBack to product

PRIVACY

Privacy, in plain language.

This page describes the behavior implemented in the SafeHuman application code. Deployment operators should keep it aligned with production configuration and any third-party services they enable.

What you submit

SafeHuman processes the link, message, email text or QR/image file that you choose to submit. Account features also process the email address you register, family invitations you initiate, identity checks you request and business integration settings you create.

What the application stores

Core scans are stored in the database even when a user is not signed in. The scan record contains a normalized preview of up to 240 characters, scan type, score, verdict, recommended action, findings and threat-intelligence metadata. The application code does not persist the full uploaded image file. Account-linked scans are associated with the signed-in user.

Uploads

Uploaded images are decoded in memory for image metadata and QR content. If a semantic vision provider is configured, the image can be sent to that configured provider for analysis. Email files can be parsed for sender/authentication metadata, attachments and embedded links.

Optional external services

When configured by the deployment operator, URL or domain data may be sent to VirusTotal, RDAP services, AbuseIPDB, URLhaus or Google Web Risk. Identity checks can send an email address to the configured breach-monitoring provider. Vision and voice analysis can send submitted media to configured provider endpoints. Those providers have their own terms and retention practices.

Model training

The SafeHuman application code contains no model-training pipeline that uses scanner submissions to train SafeHuman models. This statement does not override the independent terms of any configured third-party provider.

Authentication and account data

Passwords are stored as scrypt hashes. MFA secrets, account email, plan status, scan history, family relationships, notifications, API-key metadata and organization membership can be stored in the database as required for those features.

Retention and deletion

The current codebase does not implement an end-user self-service delete-account or configurable scan-retention workflow. Production operators should define retention policy and deletion handling before broad commercial launch. This is a documented remaining limitation rather than a privacy guarantee.

Questions or deletion requests

Use the official support/contact channel published by the operator of the SafeHuman deployment. Do not send passwords, OTPs or recovery secrets in a support request.