Decoding the world of cybersecurity

Optimizely script compromise reaches 178 organisations

A vulnerability in Optimizely Web Experimentation allowed malicious JavaScript to be added to customer projects. Researchers identified affected snippets used by 178 organisations; the company says it has removed the…

Optimizely script compromise reaches 178 organisations
Summary
  • Report URI identified 188 malicious Optimizely snippets associated with 178 organisations on 16 November.
  • Optimizely confirmed unauthorised exploitation of an Experimentation vulnerability and injection of malicious JavaScript into customer projects.
  • The vendor says the flaw has been fixed and injected code removed; site visitor exposure and actual losses vary and are not established by the snippet count.

A vulnerability in Optimizely Web Experimentation allowed an unauthorised party to insert malicious JavaScript into some customer projects, exposing visitors to websites that loaded the affected scripts. The company acknowledged the incident on its service status page and said the weakness had been fixed, malicious code removed and additional controls applied to affected accounts. Researchers at Report URI identified 188 contaminated script snippets associated with 178 organisations.

The findings were published by researcher Scott Helme on 16 November after examination of 1,611 distinct Optimizely snippets. The affected organisations included high-profile news, technology and commercial websites. The research described code capable of redirecting or manipulating browser activity, including a payment-card skimming component. Those observed capabilities demonstrate the seriousness of the injected scripts but do not establish how many visitors encountered them or whether money or data was successfully stolen.

Optimizely Web Experimentation allows site operators to test alternative page designs and content by loading scripts in visitors’ browsers. That model makes third-party JavaScript a trusted part of the user experience, often with extensive access to the page on which it executes. When an attacker can modify such code at its point of delivery, the website may serve harmful instructions despite its own application files remaining unchanged.

Report URI identified changes to snippets on 16 November and documented how the malicious code was delivered through material normally associated with the experimentation platform. The compromise reached customer browsers through a trusted third-party script, rather than requiring changes to each affected website’s own application code. A website can secure its servers and still deliver unexpected client-side functionality if a legitimate external script changes after it was approved. Standard application vulnerability scanning may not reveal a modification introduced through a dependency at runtime.

The vendor’s public update says an unauthorised party exploited a vulnerability in Experimentation to add code to some customer projects. It reports that the vulnerability was fixed, injected content removed and affected user accounts secured. These are important containment steps, although they do not provide a complete count of affected end users or establish that every potentially exposed website has finished its own review of browser activity during the incident window.

Browser security controls can limit the reach of malicious scripts, but their effectiveness depends on how a site has configured them. A Content Security Policy, for example, can restrict which origins may load scripts or receive data. Weak policies or broad allowances for dynamic script execution offer less separation. Report URI assessed some affected sites’ policies as part of its investigation, but there is no basis for assuming identical protection or exposure across all 178 organisations.

The trust problem extends beyond this particular provider. Marketing tags, analytics tools, customer experience products and payment integrations frequently run code in the same browser context as a site’s own functionality. An organisation that has approved a vendor for one purpose may unintentionally give its code access to other information on the page. That access is determined by browser execution rules and site architecture, not by the wording of the vendor contract.

Removing the malicious script does not determine which visitors received it or what browser actions followed. They may need to establish which visitors received contaminated code, what page functions were present at the time and whether any existing monitoring retained sufficient evidence. The available public research gives a list of affected snippets and examples of payload behaviour, not a definitive victim-by-victim assessment of data loss.

Optimizely says it has removed the injected code and strengthened protections against unauthorised delivery. The researchers’ findings establish a substantial browser-side exposure across customer properties, while the actual effects on individual visitors remain to be determined. Neither Report URI nor Optimizely has published a definitive total for end users whose information was taken or devices compromised.

×