Legal · Security

Verification Security Disclosure

How we protect your verification page, what we log, and what we can't promise. Last updated: May 2026.

This page exists because we want to be honest about what our security system does — and does not — do. If you've seen our homepage claims about 60-second URLs, watermarking, and capture detection, this is the technical and legal explainer. If anything here contradicts something you read elsewhere on our site, this document is the authoritative version.

1. What we collect on every scan of your verification page

When someone scans your physical service dog ID card (via QR or NFC) and your verification page loads, our server and a small client-side script collect the following data:

1a. Server-side (HTTP request)

  • IP address (used for geo-lookup; stored as a SHA-256 hash for the scan record)
  • Approximate IP-derived location: city, region (state/province), country, ISP, ASN, and approximate latitude/longitude. Source: ip-api.com, cached 24h. This is not GPS — it's an estimate from the network address, and can be wrong for VPN users, mobile carrier NAT, or corporate proxies.
  • Browser User-Agent string (the browser and OS the scanner is using)
  • HTTP Referer header (the page they came from, if any)
  • Timestamp of the scan

1b. Client-side (JavaScript)

  • Device fingerprint hash — a 32-bit non-cryptographic hash combining the signals below. Used to detect when the same device returns to scan multiple handlers (a fraud pattern).
  • Canvas signature — a hash of how the browser renders a small test image. Mildly identifying.
  • WebGL GPU renderer string — e.g. "NVIDIA GeForce GTX 1080" or "Apple M2".
  • Audio context signature — a hash of an inaudible audio buffer rendered by the browser.
  • Timezone, browser language, screen resolution and device pixel ratio.
  • Connection type (e.g. "4g", "wifi") where the browser exposes it.
  • Battery level + charging status (Chrome Android only; deprecated in most browsers — usually absent).
  • JavaScript Referrer (fallback only — server-side Referer is preferred).

1c. Capture detection events (only if triggered)

  • The detection method that fired (e.g. printscreen-key, mac-screenshot-shortcut, visibility-flicker-450ms)
  • The timestamp the detection occurred
  • A flag marking the scan as a suspected-capture event

What we do NOT collect: your name, phone number, email address, exact GPS location, camera/microphone access, contacts, files on your device, or any payment details. These would all require explicit OS-level permission prompts that we never display.

2. Why we collect it (legal basis)

UK GDPR / EU GDPR: Article 6(1)(f) — legitimate interests pursued by the data controller, namely the prevention of fraud and the protection of registered handlers from misuse of their verification page.

California Consumer Privacy Act (CCPA / CPRA): §1798.105(d)(2) — exception for detecting security incidents, protecting against malicious activity, fraud, or illegal activity.

Legitimate Interests Assessment (summary):

  • Purpose: Prevent fraudulent reuse of verification pages and provide handlers a documented forensic trail when their page is captured or saved.
  • Necessity: Without this data, a captured screenshot of a handler's page cannot be traced back to who captured it or where. Anonymous capture is the dominant fraud pattern in the US service dog registry market.
  • Proportionality: We collect only what's strictly needed for fraud detection. We do not enrich the data with third-party datasets. We do not sell, share, or use any of this data for advertising. We do not build behavioral profiles.
  • Balancing test: The data subject (the person scanning) is also the data subject most likely to be misusing the page. The registered handler (also a data subject) has a stronger interest in fraud prevention than the scanner has in anonymity for an action they performed on someone else's profile page. We make this trade-off transparent via the privacy notice on the verification page.

3. Retention

Verification scan records (including all data listed in section 1) are retained for 90 days from the scan date, then automatically deleted from our database. This window is justified as:

  • Long enough for a misuse pattern to surface (a captured screenshot rarely surfaces same-day)
  • Short enough to minimize personal-data exposure if our infrastructure is ever breached
  • Aligned with comparable fraud-detection retention windows in the financial industry (30–180 days typical)

Handlers can request earlier deletion at any time via support@adaservicedog.com.

4. What our system can and cannot do

✅ What we can do

  • Deter casual capture via a permanent on-screen warning toast that becomes part of any subsequent screenshot
  • Detect most desktop screenshot shortcuts (PrintScreen, Cmd+Shift+3/4/5, Win+Shift+S, Ctrl+P, F12)
  • Detect some mobile screenshot events (iOS visibility flicker, window blur patterns)
  • Watermark every rendered page with a unique per-scan seed and timestamp
  • Expire the verification URL after 60 seconds, so a saved screenshot cannot be used to re-verify the dog
  • Log a forensic profile of each detected capture into the handler's audit trail
  • Attempt to email the handler when a hard capture is detected (best-effort delivery)

❌ What we cannot do (be honest with yourself)

  • We cannot prevent screenshots. No website can. The screenshot tools live in the operating system, outside the reach of any web page. Only native apps (and only on some operating systems) can request screenshot suppression.
  • We cannot detect every screenshot method. External capture tools, second-camera photos of the screen, virtual machines, browser isolation, and screen-recording overlays will often evade detection.
  • We cannot guarantee handler notification. Email delivery is best-effort. SMTP bounces, spam filters, full inboxes, and rate-limit exceptions all cause undelivered alerts.
  • We cannot know exactly where the scanner is. IP-derived geo is approximate. VPNs and mobile carrier NAT often produce misleading results.
  • Our forensic data is supporting metadata, not standalone proof. If a captured screenshot is later misused, the evidentiary value of our scan log depends on chain of custody, integrity, and corroborating evidence — standards that a court would apply to any digital forensic record.
  • Anti-fingerprinting browsers (Brave, Tor, Safari, Firefox ETP) will block significant portions of our client-side collection. The system degrades gracefully — the page still works — but the audit trail will be sparser for these visitors.

5. Your rights as the registered handler

  • Access: Request a copy of all scan records associated with your registration
  • Deletion: Request immediate deletion of your scan records (we will action within 30 days)
  • Object: Object to specific processing by emailing us — we will pause or stop if your objection outweighs our legitimate interest
  • Portability: Request your scan records in machine-readable JSON format
  • Notification opt-out: Turn off capture-detected emails in your dashboard under Privacy & Alerts (we will continue logging captures to your audit trail)

6. Your rights as someone scanning a handler's page

If you scanned a service dog verification page and want to know what we collected about you:

  • Email support@adaservicedog.com with the approximate date/time of the scan and the dog's name or registration number if known
  • We will respond within 30 days under UK GDPR Article 12 or CCPA §1798.130, whichever applies to your jurisdiction
  • You may request deletion. Note: where the handler has flagged your capture as suspected misuse, we may retain the record to defend the handler's legitimate interest, in which case we will explain the retention reason

7. Contact

For all matters relating to this disclosure, verification scan data, or our security system:

Email: support@adaservicedog.com

Postal: NS Design ID Cards, 9 Coolnagarde Avenue, Omagh, BT78 1GA, United Kingdom

For broader data-protection matters, see our full Privacy Policy.

Document version: v1.0 — published 13 May 2026.

Change log: Initial publication alongside the v0.6 forensic capture system.