Safeguard
Application Security•

ASVS 5.0 V14.2.3: Keep Sensitive Data From Trackers

ASVS 5.0 V14.2.3 (Level 2): defined sensitive data must not flow to untrusted parties such as user trackers. How to find leaks and block them.

Safeguard Security Team
2 min read

ASVS 5.0 V14.2.3 is a Level 2 requirement. Data you have defined as sensitive must not be sent to untrusted parties, such as analytics and user tracking services, where you lose control over how it is collected and used.

Why it matters

Tracking scripts read page content, URLs and form values. A page address like /patients/48213/results sent to an analytics endpoint is a data transfer, even if nobody meant to send it. Once the data has left, you cannot recall it, and you may owe users or regulators an explanation.

How to implement it

Start from your classification (see V14.1.1) and keep an allowlist of third parties.

Strip sensitive parts before events leave the browser:

function safeEvent(name, props) {
  const clean = { ...props };
  delete clean.email;
  delete clean.accountId;
  clean.page = location.pathname.replace(/\/\d+/g, "/:id");
  analytics.track(name, clean);
}

Restrict what the browser may contact with a Content Security Policy:

Content-Security-Policy: default-src 'self'; connect-src 'self' https://metrics.example.com

Load trackers only on pages that hold no sensitive data, and honour consent choices before firing anything.

How to test it

Open the application with browser developer tools on the network tab, perform a sensitive journey such as viewing a record, and read every request that goes to a different host. Look for names, identifiers, tokens and full URLs in query strings and bodies. Repeat after each new third party script is added.

Common mistakes

  • Sending full page URLs that contain identifiers.
  • Session replay tools that record form fields, including masked looking ones.
  • A tag manager that lets non engineers add scripts without review.

Quick check

Does self hosting the analytics tool solve this?

It removes the third party recipient, which is the point of the requirement, but the classification rules still apply to who inside your company may see the data.

Sources

Never miss an update

Weekly insights on software supply chain security, delivered to your inbox.

Self-healing security runs on Safeguard.

Your first fix PR is minutes away.

No sales call required, even your agent can complete the purchase over MCP.