Security & Privacy
Trust is earned through transparency. Here's exactly what FraudShield AI does with your data — and what it can't do.
What may be processed
The text, sender details, links, screenshots or decoded QR destinations you choose to submit. Content is sent server-side to our AI provider for analysis. QR codes are decoded on your device and attachments are never uploaded — only their names, if you enter them.
Automatic redaction
Before analysis and storage we redact common password-like strings, one-time codes written as “code: …”, long card-like numbers and API-key patterns. Please still avoid submitting secrets.
What can be saved
With history on, the redacted text or image is stored privately in your account alongside the assessment. With history off, only the assessment is kept — not the full text or image. The assessment can still quote short, redacted snippets as evidence for its findings.
Retention & deletion
Choose automatic deletion after 7, 30, 90 or 365 days (plans also cap history length). Delete a single analysis, just its stored evidence, or all your analysis data at any time from Settings. Expired analyses are removed by an hourly cleanup. A usage count is kept for the current month so limits stay fair. You can also permanently delete your whole account from Settings.
We never open your links
Links and QR destinations are analyzed as text — domain, spelling and structure. FraudShield never visits, downloads or executes them.
Security by design
Each account's data is isolated at the database level, uploads are private, AI calls run server-side, and no provider keys reach your browser. Raw submissions aren't written to server logs. Usage limits and rate limits are enforced on our servers.
Limits of automated analysis
Results are AI-assisted and can be wrong. We don't use live threat-intelligence feeds yet, and we never claim content is definitively AI-generated. FraudShield is not an antivirus, legal service or financial advisor.
Not a guarantee
A low score doesn't mean something is safe, and a high score doesn't prove fraud. For anything involving money, codes or account access, verify independently.
How to verify suspicious content yourself
- Open the organization's official app, or type its website address yourself.
- Use contact details from your card, statement or the official site — never from the suspicious message.
- Check the sender's domain carefully for look-alike spelling.
- Never share passwords or one-time codes, and don't pay with gift cards or crypto on request.
- When in doubt, pause and ask someone you trust.