Optimizely’s Supply-Chain Attack: When a Trusted Script Becomes the Attack Surface

Published November 17, 2026

A script used for website experimentation became a route for malicious code
onto websites associated with organizations including Zoom, HSBC, CNN,
Salesforce and IKEA. The incident is a sharp reminder that a website can
serve harmful code through a trusted vendor even when the website’s own
servers have not been breached.

What happened

On November 16, security researcher Scott Helme and Report URI found
malicious JavaScript being served through Optimizely Web Experimentation.
Their analysis identified 188 infected snippets associated with 178
organizations, among 1,611 snippets they checked. That is a snapshot of
their investigation, not a confirmed count of every affected customer.

Report URI’s analysis

describes the discovery and the response as the malicious code was removed.

Optimizely said an unauthorized party exploited a vulnerability in its
Experimentation product to add JavaScript to some customer projects. The
company reported that it had fixed the vulnerability, removed the injected
code from affected projects, secured affected user accounts and added
protections against unauthorized code delivery. Its status page still
listed the security incident as unresolved while in monitoring on November
17.

Optimizely’s status page

is the place to check for further updates.

What the script was designed to do

Reporting on the malicious code says it attempted to replace cryptocurrency
addresses copied by users, capture payment-card details, steal cookies and
screenshots, and deliver exploit code targeting Apple devices. These are
reported capabilities; the full number of people affected and the extent of
any successful theft or device compromise were not known at publication
time.

Reuters’ report

also notes that the extent of resulting compromises could not immediately
be determined.

The uncomfortable lesson: “third party” still means “your page”

When a site loads a remote JavaScript file, the browser runs the code it
receives. A familiar vendor name or domain does not guarantee that the
content is safe at that moment. In this incident, the delivery chain meant
visitors could receive malicious code through a service that site owners
had added for legitimate experimentation.

That does not mean every affected organization had its internal network
breached. It does mean that a weakness in a supplier’s product can create
risk for the supplier’s customers and their visitors. Calling this a
supply-chain attack is useful, but it should not obscure the key technical
point: the browser trusted and executed changing code without an effective
check that it was still the code the site intended to run.

A fair critique of the response

Optimizely’s notice acknowledged the incident and described containment
steps. But the public update available on November 17 left important
questions unanswered: how the vulnerability worked, when unauthorized
access began, how many customer projects were affected, and what evidence
supports the claim that the injected code was removed everywhere. Those
details may emerge as the investigation continues; until then, customers
should treat the impact picture as incomplete.

Website operators also have work to do. Many organizations carefully
review server-side dependencies but allow browser-side scripts to update
dynamically with little visibility. Scott Helme argues that monitoring
script changes and deploying Content Security Policy (CSP) can improve
detection and, when suitably configured, block some unwanted behavior.

His follow-up analysis

also explains why Subresource Integrity (SRI) can be difficult to apply to
scripts designed to change frequently.

What website owners should do now

  • Check Optimizely’s incident updates and ask your Optimizely administrator
    whether your projects were affected.
  • Review the scripts your site loaded during the exposure window. Look for
    unexpected changes, new network destinations and unusual browser-side
    activity.
  • If your organization identifies affected pages, follow its incident
    response process. Consider whether payment details, sessions or other
    sensitive data entered during the exposure period require additional
    investigation.
  • Inventory third-party scripts, remove integrations that are no longer
    needed, and monitor changes to the code delivered to visitors.
  • Review CSP in report-only mode to learn what your site loads, then assess
    whether an enforced policy can safely restrict script sources and data
    destinations. Test changes carefully because experimentation tools often
    depend on dynamic behavior.

Conclusion

The Optimizely incident is a practical demonstration of how a supplier’s
access to a website’s browser can become a high-impact attack path. The
critical question for site owners is not only whether a vendor is trusted,
but how they will notice when the code delivered through that vendor
changes—and how quickly they can limit the damage.

Sources:

Report URI
;

Optimizely status
;

Reuters
;

Scott Helme’s follow-up
.

editor's pick

news via inbox

Subscribe to our newsletter and get hot new posts right into your inbox.

Leave A Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.