Optimizely supply chain attack: how one signup flaw put skimmers and iOS exploits on 178 brands' sites
The Optimizely supply chain attack put a crypto clipper, card skimmer and iOS exploits in 188 snippets. The attack chain, why it worked and what to check.
The Optimizely supply chain attack turned a website experimentation tool into a malware delivery channel for some of the most visited sites on the internet. On 16 November 2026, attackers wrote malicious code into Optimizely Web customer projects, and every page that loaded those projects’ snippets ran it. Report URI found the code in 188 of 1,611 snippets it checked, belonging to 178 organisations including Zoom, the BBC, CNN, Salesforce, GitLab, IKEA, HSBC, Equifax and Air France-KLM.
Optimizely confirmed the incident on its status page at 01:02 UTC on 17 November, saying an unauthorized party “exploited a vulnerability in Experimentation and used it to add malicious JavaScript to some customer projects”. In a statement to BleepingComputer it placed the flaw in “the Experimentation account signup flow”.
What happened
Optimizely Web works by having each customer site load a JavaScript snippet from cdn.optimizely.com. Report URI explains that the snippet contains Optimizely’s code, the customer’s experiments and an optional field called Project JavaScript (projectJS), custom code that runs on every page.
The attacker added a loader to that field and left the rest of each snippet untouched. The files were still served from Optimizely’s CDN with Optimizely’s certificate. Report URI is explicit that this was not CDN cache poisoning: the CDN served exactly what the projects contained. Optimizely’s own www.optimizely.com and login.optimizely.com snippets carried the code.
Optimizely says it has fixed the flaw, removed the injected code, revoked sessions and tokens for affected accounts and notified affected customers. It has not said how many projects were changed or how the signup flaw granted write access to other customers’ projects.
The attack chain, step by step
Based on Report URI’s timeline (all times UTC, 16 November unless noted):
- Rehearsal on other third-party scripts. From 13 November, the same actor’s injector (
webshell[.]lol/get.js) was planted through Kissmetrics, StorifyMe and Minute Media scripts. - First Optimizely injection. At 04:21 the BBC’s snippet started loading
webshell[.]lol. The payload domain1688360[.]ccwas registered at 05:23. - Crypto clipper goes live. The first build of
track.jsappeared at 08:10. It replaces valid wallet addresses (BTC, ETH and EVM chains, TRON, Solana, TON and others) with the attacker’s in page text, forms, links, the clipboard and outgoing requests. - Card skimmer added. At 13:43 the script began watching form submissions and POST requests for a Luhn-valid card number, expiry and CVV, sending them to
1688360[.]cc/card.php. Cards typed into hosted payment iframes such as Stripe or Adyen were out of reach. - Cookie and screenshot theft. When it swapped an address near payment wording, it sent the page URL, non-HttpOnly cookies and an html2canvas screenshot to
hit.php. - iOS exploit delivery. On iPhones and iPads,
boot.jsfrom two.icudomains picked an exploit chain by version: Coruna for iOS 13 to 17, DarkSword for iOS 18, and a fingerprinting-only chain for iOS 26. Cynet, quoted by The Hacker News, describes what it retrieved as conditional delivery, “not a successful compromise of a physical iPhone or iPad.” - Evasion and persistence of access. Report URI saw three loader variants: a plain script tag, a base64 URL decoded with
atob(), and an inline copy that runs viaevalif the domain is blocked. Smartsheet’s snippet was cleaned once and then re-infected with the base64 version. - Cleanup. A mass cleanup started around 20:40. By 21:52 only one snippet (TalkTalk’s) was still infected. By 19:25 the attacker’s wallets had received about $7,040, though some wallets predate the campaign, so not every transfer is a victim’s.
Why it worked
Trusted origin, untrusted content. Every affected site had allowlisted cdn.optimizely.com in practice and often in its Content Security Policy. Report URI notes that allowing that host did not help, because the code came from it.
A low-privilege entry point reached high-privilege data. By Optimizely’s account, the way in was the account signup flow. Whatever the exact bug, a path available to new accounts ended with write access to existing customers’ project code. That is a tenant isolation failure, not a CDN problem.
Custom code fields are code deployment. Project JavaScript is a production deployment channel to every page, but it rarely gets the review or monitoring a normal release would.
Third-party tags on sensitive pages. The skimmer only worked where card fields lived in the site’s own DOM. Sites that kept card entry inside the payment provider’s cross-origin iframe were out of reach of that part of the payload.
What defenders should do
- Inventory third-party JavaScript, who can change it and which pages it reaches. Marketing tags count.
- Keep third-party tags off checkout and login pages. Take card data only through the payment provider’s hosted fields.
- Use a strict CSP with tight
script-srcandconnect-src, and avoidunsafe-eval. A blockedconnect-srcstops exfiltration even when a trusted host serves bad code. - Monitor client-side changes. Hash or diff the vendor scripts you serve and alert on new script loads and new outbound destinations from the browser.
- Treat vendor consoles as production access: SSO, MFA, least-privilege roles and audit logs for anyone who can edit custom code.
How to check if you are affected
- Your snippets: open Project JavaScript in every Optimizely project and the custom code in every experiment. Search for
1688360,atob(,track.js,.icuor any script you did not write. A snippet that is clean now may still have served the code on 16 November. - Network IOCs:
1688360[.]cc(paths/track.js,/card.php,/hit.php),136.113.106[.]136,5vobauua3ojpklx[.]icu,eil786npto9u5bq[.]icu,webshell[.]lol. StepSecurity, cited by The Hacker News, advises searching logs back to 08:00 UTC on 16 November. - Hashes:
track.jschanged many times, so one hash will not catch every version. The Hacker News lists the earliest and final SHA-256 values. - Ask Optimizely which of your project IDs were changed, when and how, and review which users, API tokens and integrations can write to your account. Rotate them.
- Exposure: if your own forms take card data, treat cards entered on 16 November as possibly exposed. Treat secrets in non-HttpOnly cookies as exposed and rotate them.
- Visitors: update iPhones and iPads to iOS 18.7.3 or later, or 26.3 or later, and compare any crypto payment made that day with the intended address.
My take
This is the same pattern as the Cursor supply chain breach: the attacker did not need to breach CNN or Zoom, only a vendor every one of them already trusted. The difference is scale. One flaw in one SaaS signup flow reached 178 organisations in a single day, and none of them changed a line of their own code.
What bothers me most is that the defences many teams rely on assume the trusted host stays trustworthy. Domain allowlists and “it comes from our vendor’s CDN” failed here at once. For more cases where trust in a vendor became the attack path, see the supply chain category.
Three questions worth asking this week: which third-party scripts run on our checkout and login pages, who outside engineering can push code to every page through a vendor console, and would we notice if one of those scripts started talking to a new domain?
Sources
- Web Experimentation Security Incident · Optimizely status page, 17 November 2026
- World's Largest Brands Infected With Malware Via Optimizely · Report URI (Scott Helme), 16 November 2026
- Optimizely hack infected CNN, Fox News, BBC, Salesforce sites with malware · BleepingComputer, 16 November 2026
- Hijacked Optimizely Scripts Served Card Skimmer and iOS Exploit Kits to CNN, BBC and Zoom Visitors · The Hacker News, 17 November 2026