Skip to main content
How secure are your AI vendors? We assessed 18,160 of them. Read the 2026 AI Supply Chain Report.Read the Report
BlackKite: Home
Menu
blog

Optimizely: Supply-Chain Attack on Experimentation Platform Delivers Malware to 178 Organizations

Published

Nov 18, 2026

Authors

Ferdi Gül

IN A NUTSHELL

On November 16, 2026, an unauthorized party exploited a vulnerability in the account signup flow of Optimizely's Experimentation (Optimizely Web) product to inject malicious JavaScript into the "Project JavaScript" field of customer projects. Because the tampered files continued to be served from Optimizely's own trusted CDN (cdn.optimizely.com) under its valid certificate, the code reached visitors of at least 178 organizations, including Zoom, CNN, Fox News, the BBC, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger, and Peacock, enabling cryptocurrency wallet-address swapping, payment card skimming, cookie and screenshot theft, and delivery of the Coruna and DarkSword iOS exploit kits to iPhone and iPad visitors. No CVE, CVSS score, or CWE has been assigned because the flaw sits in Optimizely's own signup workflow rather than a versioned product. Optimizely has confirmed the incident, blocked the vulnerable path, deployed a fix, removed injected code, ended sessions, revoked tokens, and reset access for affected accounts; exploitation is confirmed and was active for roughly 16-19 hours on November 16 before cleanup, with no indication it continues.

Key Takeaways

  • On November 16, 2026, Optimizely confirmed (November 17) that an attacker exploited a vulnerability in its Experimentation account signup flow to inject malicious JavaScript into customer projects.
  • Security researcher Scott Helme of Report URI found the malicious code in 188 of 1,611 Optimizely snippets checked, affecting at least 178 organizations including CNN, the BBC, Zoom, Salesforce, GitLab, HSBC, Equifax, and Ledger, with the real number believed to be higher.
  • The injected code was served from Optimizely's own trusted CDN under its valid certificate, so listing the Optimizely domain in a Content Security Policy offered no protection because the malware was delivered from that same domain.
  • The payload swapped cryptocurrency wallet addresses, skimmed payment card data from forms not served by hosted iframes, stole non-HttpOnly cookies and screenshots, and on iPhones and iPads attempted to deliver the Coruna and DarkSword iOS exploit kits.
  • No CVE, CVSS score, or fixed version number applies because this was a SaaS platform flaw in Optimizely's own infrastructure; Optimizely says it has fixed the flaw and removed all injected code but has not disclosed the exploit mechanism or full scope of affected projects.

Executive Summary: A Trusted Third-Party Script Became the Attack Surface

A supply-chain compromise at Optimizely, a widely embedded digital-experience and A/B-testing platform, served malicious JavaScript to visitors of at least 178 organizations on November 16, 2026, through a script that every affected website had already chosen to trust. No server at any of the affected organizations was breached; the code arrived through a JavaScript snippet those organizations load from Optimizely's own content delivery network. This matters for third-party risk teams because it demonstrates that a vendor's own trust boundary, not a customer's infrastructure, can become the attack surface, and because the affected list includes banks, airlines, telecoms, news outlets, and SaaS platforms across sectors.

Key findings: A Signup-Flow Flaw Let Attackers Edit Customer JavaScript

  • An unauthorized party exploited a vulnerability in Optimizely's Experimentation account signup flow to gain the ability to write malicious code into the "Project JavaScript" field of customer projects, a field that executes on every page view.
  • The tampered files were still served from cdn.optimizely.com with Optimizely's valid certificate, so browsers and most Content Security Policies treated the code as trusted.
  • The payload combined cryptocurrency address swapping, payment card skimming, cookie and screenshot theft, and iOS exploit-kit delivery (Coruna and DarkSword) in a single multi-purpose script.
  • Optimizely has confirmed the incident, fixed the vulnerable path, removed injected code, and secured affected accounts, but has not disclosed how the signup-flow flaw granted write access across unrelated customer projects.

How the Optimizely Attack Was Discovered and Timeline of Events

The compromise was discovered and documented by security researcher Scott Helme of Report URI, who monitors third-party JavaScript changes on customer sites. Optimizely confirmed the incident publicly after being contacted by BleepingComputer and posted a status notice. Separate researchers, including Hudson Rock, StepSecurity, and Kirk of Ohey, extended the investigation to related infrastructure on other vendor platforms.

NOTE  Researchers have linked scripts from Kissmetrics, StorifyMe, and minute-ly to the same loader code and infrastructure, with tampered responses recorded as early as November 13. Shared infrastructure does not confirm a single operator.

Time (UTC)

Event

Sep 17, 2025

The same actor's injector, webshell[.]lol, first appears on Turkish sites via an SEO "backlink panel" script, per Report URI.

Nov 09, 17:52

5vobauua3ojpklx[.]icu, an iOS exploit-kit router domain, is registered.

Nov 13

Earlier wave: webshell[.]lol/get.js injected through Kissmetrics, StorifyMe, and minute-ly scripts on sites including securitybank.com, faz.net, smh.com.au, and foxsports.com, per Report URI and Hudson Rock.

Nov 16, 02:21

Last clean urlscan capture of the BBC's Optimizely snippet.

Nov 16, 04:21

First Optimizely injection observed, in the BBC's snippet.

Nov 16, 05:23

The payload domain 1688360[.]cc is registered.

Nov 16, 08:10

First build of the track.js crypto clipper goes live (rebuilt 26 times that day).

Nov 16, 13:03

Attacker sweeps stolen crypto to exchanges.

Nov 16, 13:43

Card skimmer added to track.js.

Nov 16, ~17:13-17:57

Second iOS router domain added; Coruna iframe injected into every iOS 18.6-or-earlier visit.

Nov 16, ~20:40-21:10

Mass cleanup begins (Ledger 20:41, Smartsheet 20:56, BBC 21:07, per Report URI).

Nov 16, 21:52

One of 188 infected snippets remains (TalkTalk).

Nov 17, 01:02

Optimizely posts its incident notice, confirming the signup-flow vulnerability.

Nov 17, 00:37

Registry places 1688360[.]cc on hold, per StepSecurity; domain no longer resolves.

Nov 17, 04:11-05:39

Attacker empties the LTC, ETH, BTC and TRON wallets, per StepSecurity.

Technical Details: How Malicious Code Spread Through Optimizely's CDN

Optimizely Web Experimentation works by having each customer load a JavaScript "snippet" from Optimizely's CDN, for example at a URL such as cdn.optimizely.com/js/<project-id>.js. The snippet contains Optimizely's own code, the customer's configured experiments, and an optional field called Project JavaScript (projectJS): custom code the customer writes in the Optimizely dashboard, which runs on every page load.

According to Report URI's analysis, the attacker exploited a flaw in the Experimentation account signup flow to gain write access and inserted a loader directly into the Project JavaScript field of affected customer projects, leaving the rest of each snippet untouched. Because the modified file was still served from cdn.optimizely.com under Optimizely's own certificate, it carried the full trust of a host every affected site had already allow-listed. This was not CDN cache poisoning: the CDN served exactly what the project configuration contained, per Report URI and StepSecurity. Customers who serve their snippet from their own domain, such as opt.scheels.com or www.vodafone.com.au, were affected as well, since the content still originates from the same Optimizely project.

Three loader variants were documented:

  • A plain <script> tag pointing to https://1688360[.]cc/track.js, used in most infected snippets.
  • The same URL hidden in Base64 and decoded with atob(), which defeats a plain-text search for the domain string.
  • A full inline copy of the payload executed through eval, used as a fallback if the primary domain was blocked; this variant ran even when the main payload file was blocked on at least one HSBC site.

Once loaded, the track.js payload (roughly 30 KB, rebuilt 26 times during the day) combined four capabilities:

  • Cryptocurrency address swapping: continuously scanned page text, form fields, links, images, the clipboard (including navigator.clipboard and navigator.share), and outgoing fetch/XHR request bodies for wallet addresses across BTC, ETH and other EVM chains, TRON, Solana, TON, Litecoin, Bitcoin Cash, and Dogecoin, validating checksums and substituting attacker-controlled addresses only near payment-related words such as "wallet," "deposit," or "send."
  • Payment card skimming: added at 13:43 UTC, hooked form submissions to capture card numbers, expiry dates, and CVV codes entered into a site's own form fields and transmitted them to https://1688360[.]cc/card.php; it could not read card fields inside cross-origin hosted payment iframes such as Stripe or Adyen.
  • Data exfiltration: under certain conditions, collected the page URL, JavaScript-readable (non-HttpOnly) cookies, and a rendered screenshot of the page.
  • iOS exploit delivery: on detecting an iPhone, iPad, or iPod, loaded a secondary script (boot.js) from .icu domains that fingerprinted the iOS version and routed the device, in a hidden iframe, to the Coruna exploit kit (targeting iOS 13.0-17.2.1, per Google Threat Intelligence Group) or the DarkSword exploit kit (targeting iOS 18.4-18.7, per iVerify), while iOS 26 devices triggered a chain that so far only fingerprints the device. Affected pages on iOS 18.6 or earlier reloaded every 60 seconds between 17:13 and 17:57 UTC to retry the exploit, a behavior visitors might have noticed.

NOTE No public Proof-of-Concept exploit code has been published for this incident; the mechanics above are drawn from vendor and researcher technical write-ups rather than released exploit code.

Affected Versions: No CVE or Patch Because This Is a SaaS Flaw

This is a cloud-delivered SaaS platform vulnerability in Optimizely's Experimentation (Optimizely Web) product rather than a versioned on-premises package, so no discrete affected version range or fixed version number applies. Any organization embedding the Optimizely Web JavaScript snippet between the first observed injection in the early morning of November 16, 2026, and the completion of cleanup later that day was potentially affected. StepSecurity estimates the exposure window at about 16 hours for most affected sites.

Optimizely states it has fixed the underlying signup-flow vulnerability platform-wide and removed all injected code; Report URI found only 1 of 188 infected snippets remaining at 21:52 UTC on November 16, with every file reported clean by 23:11 UTC the same day. Separate but related injection campaigns using similar loader and hosting infrastructure were traced by Hudson Rock and researcher Kirk of Ohey to three other third-party services, Kissmetrics, StorifyMe, and minute-ly, with tampered responses recorded as early as November 13; shared infrastructure does not confirm a single operator or an identical entry point into those services.

Who Is Affected: Banks, Airlines, Media and SaaS Companies Using Optimizely Web

Any organization that embeds an Optimizely Web Experimentation JavaScript snippet on a public-facing website was potentially exposed, regardless of the organization's own security posture, since no customer server was breached. Report URI's review of 1,611 Optimizely snippets found 188 carrying malicious code, belonging to 178 organizations (some organizations run more than one project), and believes the real number is higher because only snippets it could locate in time were checked.

Per Report URI's sector breakdown, banking and finance was the hardest-hit sector with 32 organizations, followed by technology and SaaS with 23, retail and fashion with 20, streaming with 12, news and media with 10, airlines and travel with 9, telecoms with 7, crypto and trading with 6, and insurance with 6. Named organizations across sources include Zoom, CNN, Fox News, the BBC, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger, Peacock, ASOS, Alaska Airlines/Hawaiian Airlines, Virgin Media, Virgin Media O2, Vodafone Australia, O2, Mutual of Omaha, American Eagle, American Girl, Scheels, Stamps.com, Smartsheet, Hootsuite, Change.org, Bitly, IG, UNICEF USA, and TalkTalk. Optimizely's own public-facing sites, www.optimizely.com and login.optimizely.com, also carried the injected code.

Salesforce confirmed it is investigating unauthorized Optimizely script modifications and has disabled Optimizely scripts across preview and production instances as a precaution. As of this writing, Zoom, Smartsheet, and GitLab had not published separate public statements on the incident.

How to Detect the Optimizely Malware: Signals and Log Searches

Organizations that embed the Optimizely Web snippet should look for the following signals, per Report URI and corroborating researchers:

  • Unexpected or unrecognized content in the Project JavaScript field of any Optimizely project, or edits to that field around November 16, 2026.
  • Network requests to 1688360[.]cc, webshell[.]lol, or 5vobauua3ojpklx[.]icu in DNS, proxy, WAF, or CSP violation-report logs.
  • Page source containing atob( calls or inline eval blocks, since one loader variant hid the malicious URL in Base64 and another ran a full backup payload via eval.
  • Repeated page reloads approximately every 60 seconds on iOS devices running iOS 18.6 or earlier, which indicates a retrying exploit attempt.
  • Re-infection after cleanup; Smartsheet's snippet was cleaned once and then re-infected with a hidden loader variant, which Report URI says indicates the attacker retained access at that point.

NOTE A clean scan today does not prove a site was not affected, since the injected code has already been removed platform-wide; organizations should review historical logs back to November 16, and ideally November 13, rather than relying on a current scan.

Mitigation and Patching: Audit Scripts, Reset Access and Review Logs Back to November 13

  1. Review the Project JavaScript field and all custom experiment code in every Optimizely Web project; remove anything not recognized, and check the change history for edits around November 16, 2026.
  2. If the snippet's contents cannot be confirmed clean, disable it until it can be verified. (Partial, interim measure.)
  3. Reset Optimizely account passwords, review user accounts and API tokens, and enforce MFA or SSO on Optimizely accounts.
  4. Search DNS, proxy, WAF, and CSP logs for 1688360[.]cc, webshell[.]lol, and 5vobauua3ojpklx[.]icu; also search page source for atob( and inline eval blocks, since a plain domain-string search can miss the Base64-hidden variant.
  5. Extend log review back to November 13, 2026, not only November 16, since the same actor's loader code also appeared through Kissmetrics, StorifyMe, and minute-ly scripts starting that date.
  6. Treat any payment card entered on an affected site on November 16, 2026 as possibly exposed; inform payment processors and follow applicable PCI and legal notification obligations.
  7. Treat any secret stored in a cookie without the HttpOnly flag as exposed; revoke the associated sessions and tokens.
  8. If the snippet is served from your own domain or proxy, purge caches and replace any locally cached copies; infected snippets may also persist in users' browsers.
  9. Add a strict Content Security Policy covering script-src and connect-src, starting in report-only mode; most affected sites had no CSP or one too weak to block the payload. (Partial, reduces but does not eliminate exposure since the Optimizely domain itself was the delivery vector.)
  10. Do not rely on Subresource Integrity (SRI) for the Optimizely snippet, since it changes whenever experiments change and a fixed hash will not hold; instead monitor the served file for unexpected changes.
  11. Maintain an inventory of all third-party scripts running on your pages and alert on unexpected changes.
  12. Advise iOS users to update to iOS 18.7.3 or later, or iOS 26.3 or later, which address the vulnerabilities exploited by the Coruna and DarkSword exploit kits.
  13. Identify third-party vendors that use Optimizely Web and request written confirmation of impact and cleanup.
  14. Follow the Optimizely incident status page until the incident is marked resolved and a root cause is published; as of this writing the root cause of the signup-flow exploitation is not known.

Optimizely Attack Indicators of Compromise (IOCs): Domains, Files and Code Patterns

Type

Value

Context

Domain

1688360[.]cc

Primary payload domain hosting track.js (crypto clipper, card skimmer) and card.php (card exfiltration endpoint); registered Nov 16, 2026 at 05:23 UTC; placed on hold by registry Nov 17 at 00:37 UTC.

Domain

webshell[.]lol

Earlier-stage injector domain, same actor, first seen Sep 17, 2025 on Turkish sites and again in the Nov 13 Kissmetrics/StorifyMe/minute-ly wave.

Domain

5vobauua3ojpklx[.]icu

iOS exploit-kit router domain, registered Nov 9, 2026 at 17:52 UTC.

Domain

eil786npto9u5bq[.]icu

Second iOS exploit-kit router domain, registered Nov 16, 2026 at 16:38 UTC.

File

boot.js

Script loaded on iOS devices from the .icu router domains that fingerprints iOS version and opens the Coruna/DarkSword exploit chain in a hidden iframe.

Code pattern

atob( / inline eval

Marks the Base64-obscured loader variant and the backup payload variant; a plain-text domain search will miss these.

Managing Third-Party Risk From Embedded Scripts

This incident reached third parties through a dependency, not a breach of their own infrastructure: every affected organization trusted a script it did not write and could not have inspected for a malicious change in real time. That is the core third-party risk exposure Black Kite tracks for this kind of event, and it is why Black Kite has published a FocusTag, "Optimizely Supply Chain Attack," to identify vendors in customer portfolios that load the Optimizely Web snippet and may have served the malicious payload to their own site visitors.

Because there is no CVE or patch version to check, portfolio assessment for this incident depends on identifying which vendors embed the Optimizely Web snippet (through asset and script inventory rather than a version banner) and following up directly with those vendors on remediation status, consistent with the "Recommended Actions" item to request written confirmation of impact and cleanup from any vendor using Optimizely Web.

References

About the author

Ferdi Gül · Vulnerability Research Strategist

Ferdi Gül is Vulnerability Research Strategist at Black Kite, leading development of FocusTag and authoring the weekly Focus Friday vulnerability series for third-party risk teams.