deniz.in

Markets

Weather

Loading weather

· via dev.to (home feed)

Optimizely snippet hijack delivered card skimmers and iOS exploits to 178 organizations

Attackers exploited a flaw in Optimizely Web Experimentation to plant a malicious loader in customer snippets, exposing visitors of 178 organizations to crypto clippers, card skimmers and iOS exploit kits.

Optimizely snippet hijack delivered card skimmers and iOS exploits to 178 organizations

What happened

For roughly 17 hours on November 16, the JavaScript snippet that Optimizely Web Experimentation serves to customer websites became a delivery channel for financial theft and iPhone exploitation. According to a detailed write-up published on dev.to, an attacker exploited a vulnerability in Optimizely's Experimentation product and injected a malicious loader into the Project JavaScript field of customer projects. Every page that loaded the official snippet from cdn.optimizely.com — signed by the vendor's own certificate — then executed that loader.

Report URI, which monitors the scripts websites run, found the injected code in 188 of the 1,611 Optimizely snippets it examined, belonging to 178 organizations. The affected list includes CNN, the BBC, Zoom, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger and Peacock, along with Optimizely's own main site and login page. Some researchers call the campaign OPTIMJACK, and Report URI notes the real count is probably higher because it could only check the snippets it found in time.

How the loader got in

The Optimizely snippet refreshes itself whenever a customer changes settings and runs the custom Project JavaScript field on every page load. The attacker wrote a loader into that field and left the rest of the snippet untouched, so the malicious code traveled the vendor's normal delivery path. Sites serving the snippet from their own domains were affected too, because the content comes from the same project.

The loader appeared in three forms: most fetched a script from a hard-coded domain, one concealed that domain with base64 encoding, and a third carried an embedded copy of the payload so it kept working even if the domain was blocked. Cynet, whose endpoint detections flagged browser connections to the attacker's server, traced the loader back to a file on cdn.optimizely.com and captured the code running on CNN and Zoom pages. Serving the code does not mean those sites' own systems were breached.

What the payload did

A payload analysis by StepSecurity describes code built around financial theft. It replaces legitimate cryptocurrency wallet addresses with the attacker's own in page text, forms, links, the clipboard and outgoing requests, but only when an address passes its chain's checksum and sits near payment wording such as wallet, deposit or send — a filter that makes the swap harder to notice.

At 13:43 UTC a card skimmer was added, capturing card numbers, expiry dates and CVVs from the site's own form fields and forwarding them to the attacker's server. There are limits: it cannot read cards entered in cross-origin payment iframes such as Stripe's or Adyen's hosted fields, and it cannot touch cookies marked HttpOnly. After swapping an address near payment wording, it also sent the visitor's cookies and a page screenshot to the attacker. By 19:25 UTC the attacker's wallets held about $7,040, though Report URI could not determine how many card numbers were exposed.

The iOS exploit chain

On iPhones and iPads the code fetched a second-stage script that picked an exploit kit by iOS version: Coruna for iOS 13.0 to 17.2.1 and DarkSword for iOS 18.4 to 18.7, both previously documented by Google researchers. For iOS 26 a separate chain so far only collects device details. Between 17:13 and 17:57 UTC the attacker escalated, forcing a Coruna iframe onto every visit from iOS 18.6 or earlier, with the page reloading every 60 seconds to retry exploitation. Cynet retrieved that stage using an emulated iPhone browser and describes it as delivery of the exploit rather than a demonstrated takeover of a physical device. No confirmed device compromises have been reported.

Cleanup and open questions

Mass cleanup began around 20:40 UTC, and by 21:52 UTC only TalkTalk's snippet still carried the code. Optimizely posted an incident notice at 01:02 UTC on November 17 saying it had fixed the flaw, removed the code, secured affected user accounts and added protections against unauthorized code delivery. The notice does not explain the root cause, say how many projects were modified, state when the fix went in, or confirm the attacker's access is closed. Smartsheet's snippet was cleaned once and then re-infected with an obfuscated loader, which Report URI says indicates the attacker still had write access at that point.

Why it matters

This is the third-party script problem demonstrated at full scale. One flaw in a vendor product gave the attacker a write path into 178 organizations, and the malicious code arrived over a trusted CDN with a valid certificate, so content security policies and trusted-domain lists offered no defense. The actor had already run the same playbook through Kissmetrics, StorifyMe and Minute Media scripts, after first appearing on Turkish sites in September, which suggests a systematic search for writable script supply chains rather than a one-off. For organizations running Optimizely Experimentation snippets, the exposure window and any regulatory obligations around card data remain open questions until the vendor discloses the root cause and confirms attacker access is fully closed — and until then, a single clean check of a snippet is not proof it stays clean.

  • #security
  • #supply-chain
  • #javascript
  • #malware
  • #card-skimming

Related posts