Fixing One Vendor Does Not Fix Your Website - Source Defense
RESOURCES / BLOG

Fixing One Vendor Does Not Fix Your Website

Every customer-facing website that runs browser-side JavaScript has a persistent control flaw. Code can be approved, delivered through a trusted source, and still read data, alter a page, load more code, or transmit information outside the website owner’s authorization. Removing one compromised vendor script does not fix that flaw. It only closes one route into it.

The feeling that everything is broken has gone global. The loud failures make headlines. Websites have a quieter failure mode: they can appear to work normally while code inside the browser acts outside the owner’s control.

Optimizely has confirmed that an unauthorized party exploited a vulnerability in its Experimentation product and used it to add malicious JavaScript to some customer projects. The company says it fixed the vulnerability, removed injected code, secured affected accounts, and added protections against further unauthorized code delivery.

Ars Technica, citing Report URI’s investigation, reported that the campaign included cryptocurrency-wallet address swapping, payment-card skimming, cookie and screenshot theft, and iOS exploit-kit delivery.

Those actions close an attack path. They do not resolve the broader browser-side control gap that made the attack consequential: trusted code can change, and many website owners do not retain control over what that code does once it reaches a customer’s browser.

The story is bigger than one vendor

A Beazley Security advisory reported malicious third-party scripts connected to this activity through Optimizely, Kissmetrics, StorifyMe, and minute-ly. The available public research does not establish a single shared root cause across those providers. It does reveal a shared security condition: trusted third-party code can become a delivery route for unauthorized browser behavior.

A vendor patch can remove one known route. A vendor offboarding can remove another. Neither action gives your team control over the code that still runs across the customer journey through experimentation tools, analytics, tag managers, chat services, payment technologies, support platforms, and internal deployments.

Trust is an approval decision not an operating control

Vendor reviews, security questionnaires, code reviews, and source allowlists matter. They help determine what your organization is willing to deploy. They do not determine what a script is allowed to do in every live customer session.

Once browser-side JavaScript loads, it may interact with page elements, user input, cookies, storage, network connections, and additional code. That authority is what makes modern websites work. It is also what creates exposure when an approved script changes, an inherited dependency is compromised, or a configuration is altered without your team’s knowledge.

Third-party tools belong on modern websites. The operating question is whether your team can control their behavior after they arrive.

Detection cannot stop behavior already in the browser

An alert tells you that a script has crossed a line. It cannot undo data access, reverse a page change, or stop behavior that occurred before a human investigated and applied a policy. A tool that only inventories or alerts on scripts cannot control what those scripts do in a live customer session.

Content Security Policy and Subresource Integrity also have useful roles in a layered program. They can constrain loading sources, execution methods, and known resources. They cannot independently decide whether code delivered through an approved source may read a password field, alter a checkout page, access browser storage, or transmit data to an unauthorized destination.

Website control has to extend across the customer journey

Checkout pages deserve protection. So do login, registration, password-recovery, onboarding, account, loyalty, booking, and lead-generation pages.

A script does not need to capture a card number to cause material harm. Credentials, personal data, session information, and page controls can all be valuable targets. Security programs need to follow the data and customer journey, rather than stopping at the boundary of one payment page or one vendor relationship.

The Source Defense Platform for Web Control gives teams visibility into browser-side JavaScript behavior and, where their selected plan and policy configuration support enforcement, applies runtime controls over what code can access, alter, and send. Teams can use behavior-based policies, including redaction, isolation, and blocking, to establish boundaries before the next alert arrives.

AI makes point in time approval less durable

The PCI Security Standards Council has warned that AI-powered vulnerability discovery increases the importance of frequent, even continual, monitoring and management. AI will shorten the cycle between payload creation, variation, and deployment. A point-in-time approval model becomes less durable when attackers can rebuild and tailor browser-side code faster than teams can review and react.

Start with the code you already trust

A practical website-control program begins with four questions:

  • Which scripts and inherited dependencies run on every customer-facing page?
  • What sensitive data, browser storage, and page elements can each script access?
  • Which destinations can each script communicate with?
  • What control takes effect when behavior moves outside its approved purpose?

The next attack may use a different vendor, payload, or delivery path. The control gap will be the same.

Do not wait for the next vendor disclosure. Control what browser-side code can access, alter, and send now.

Related Reads

Source Defense
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.