Illustration of a browser window with a purple puzzle piece fitted at the top, from which a magenta hook hangs and snags a navy payment card.

Optimizely Served Malicious Code to 178 Organizations

A
Admin
·

On 16 November 2026, the Optimizely Web script, an A/B testing tool, served malicious code to visitors of 178 organizations, including Zoom, the BBC and HSBC, according to security firm Report URI. The code tried to steal payment cards and cryptocurrency. Any website that loads third-party scripts shares the same weak point.

What we know

  • Who found it: researcher Scott Helme of Report URI, who published his findings on 16 November at 22:47 UTC. Report URI sells script-monitoring tools, and the post mentions them.
  • How it got in: according to Report URI, the attacker wrote code into the "Project JavaScript" field of customer projects, a slot for custom code that runs on every page. The file was still served from Optimizely's own servers, with its certificate. Report URI rules out CDN cache poisoning.
  • What Optimizely says: according to Gizmodo, a spokesperson said an unauthorized party exploited a vulnerability in the Experimentation account signup flow, and that the company blocked that path and deployed a fix. BNO News quotes the company saying the vulnerability is fixed and the injected code removed.
  • How far it reached: Report URI checked 1,611 Optimizely snippets and found malicious code in 188, belonging to 178 organizations. It warns the real number may be higher. It gives no count of affected visitors.
  • When: the first injection it saw was at 04:21 UTC on 16 November, in the BBC's snippet. By 21:52 UTC only one snippet was still infected.
  • What the code did: it swapped crypto wallet addresses for the attacker's; from 13:43 UTC it copied card number, expiry and security code typed into a site's own forms; it sent cookies and screenshots; and on iPhones and iPads running older iOS versions it loaded exploit kits. Cards typed into hosted payment iframes from Stripe, Adyen and similar providers were out of its reach.
  • What it took: about $7,040 in cryptocurrency by 19:25 UTC on 16 November, according to Report URI.
  • What stopped it: among the affected organizations whose Content Security Policy (CSP) it reviewed, 68 had none and only 24 blocked the code. Allowlisting Optimizely's domain did not help, because the code came from there.

What changes and what doesn't

  • The affected sites were not hacked themselves. The code arrived inside a script from a vendor they trusted.
  • You don't need to use Optimizely for this to matter. A/B testing, analytics, chat widgets, ad pixels and maps are all third-party scripts that run with the same power as a page's own code.
  • The vendor says it is fixed. According to Optimizely, the entry path is closed. Report URI says it does not know how the attacker got write access to so many unrelated projects.
  • The useful question is not "do I use Optimizely?" but what would happen if any script your site loads were hijacked tomorrow.

How to tell if it affects you

  1. If you use Optimizely Web: search your Project JavaScript and experiment code for the domain 1688360.cc, the function atob( or any script you don't recognize. Review who has access to the account, rotate passwords and API tokens, and treat cards entered on your own forms on 16 November as exposed.
  2. List your third-party scripts. Open the browser's developer tools (F12), go to the Network tab, filter by JS and reload your home page and your checkout page. Every domain that isn't yours is a vendor that can run code on your site. Remove the ones you no longer use: a forgotten pixel from an old campaign is risk with no benefit.
  3. Move card entry off your own form. If your store takes card details in its own form, switch to the hosted field or hosted page from your payment provider (Stripe, PayPal, Adyen or another). In this attack, cards typed into those iframes were out of reach.
  4. Add a CSP and keep connect-src tight. A Content Security Policy tells the browser which domains a page may send data to. A short connect-src list stops stolen data from leaving even if a malicious script loads. Start in report-only mode (Content-Security-Policy-Report-Only) so nothing breaks, and avoid 'unsafe-eval'.
  5. Guard marketing tools like your site's admin panel. Google Tag Manager, A/B testing and chat accounts can inject code into every page: use two-step verification, limit access to the people who need it and review the change history.
  6. Set HttpOnly on session cookies. The code stole cookies that JavaScript can read. Session cookies should carry the HttpOnly flag; in developer tools, Application → Cookies shows which ones do.

Related: Cursor Breach and Fake Claude Code Extension: What to Check

Sources

Updates: any changes backed by a primary source will be added here.