Attackers abused a flaw in Optimizely's Experimentation account signup flow on November 16 to plant malicious JavaScript in 188 customer script files across 178 organizations, and sites including Zoom, CNN, GitLab and Salesforce served it to visitors. The code was built to skim payment cards, swap crypto wallet addresses and push iOS exploit kits. Optimizely says the hole is closed and the code removed. The Optimizely malware check on your own site is still your job: the vendor cleaned its files, not your orders, logs or customer messages from that day.
Key takeaways
- Attackers abused Optimizely's Experimentation signup flow on November 16 to inject JavaScript into 188 script files across 178 organizations.
- The payload targeted card data and crypto wallet addresses, collected cookies and screenshots, and loaded the Coruna and DarkSword iOS exploit kits.
- By 21:52 UTC only one of the 188 files was still malicious. Nobody has named a CVE, an affected version or an attacker.
- The vendor's cleanup does not cover your orders, your PCI DSS 6.4.3 and 11.6.1 duties or your customer notices.
- Expect clean server logs. Check DNS, gateway and CSP telemetry instead.
Was your Optimizely project one of the 188 files?
Check, don’t guess. Scott Helme first reported the campaign on Monday. He says his team analyzed 1,611 Optimizely scripts and found 188 infected files across 178 organizations. That is about 11.7% of the scripts they examined, not of every Optimizely customer.
The window was short. Helme’s earliest sighting came in the early morning of November 16. By 21:52 UTC, only one of the 188 files still carried malicious code. Nobody has said how long the flaw existed or how many visitors got a poisoned file.
There is no CVE, no affected version, no named attacker and nothing for you to install. A snippet that still runs your A/B tests proves nothing either. Optimizely says the attacker added code to projects, so it could sit right beside working experiment logic.
What should you check before you tell customers anything?
Optimizely gave BleepingComputer a list of what it did. It blocked the vulnerable path and deployed a fix. It removed the injected code, ended sessions, revoked tokens and reset access for affected users. It also blocked code delivery through Cloudflare. The company says it notified all affected customers.
That covers the vendor’s side of the fence. Your own cleanup needs at least one pass on day one:
- Check the notice and the history. Look for Optimizely’s message. Then review November 16 project changes that nobody on your team made.
- Save evidence before purging caches. Capture the JavaScript your pages served, plus response headers, CDN cache status and UTC timestamps. A cleaned project does not show what browsers received that morning.
- Fetch from more than one place. One clean fetch can miss a stale cached copy in another region.
- Rotate your side. API keys and integrations that talk to Optimizely deserve the same treatment Optimizely gave its tokens.
- Map the pages. Flag checkout, login and anything that displays a wallet address.
Salesforce took the blunt approach. It disabled Optimizely scripts across preview and production while it investigates. If you cannot answer the list above today, pull the tag from payment and login pages for a few days. That is a reasonable precaution.
Don’t treat the brand list as the scope. BleepingComputer confirmed loads through urlscan.io records. Those are snapshots of single page loads, and they say nothing about a small shop whose project was edited too.
From our engagements: after a script incident, server logs come back clean, and that is expected. A browser skimmer sends data straight from the visitor’s device to the attacker. Your server only sees the request for the vendor file. The evidence sits in DNS logs, web gateway records and CSP violation reports. If you collect none of those, centralized log monitoring is the gap.
Cards, wallet addresses and sessions: what a visitor could lose
The injected loader fetched a second script, track.js, from an attacker-controlled domain. Some copies used the URL directly. Others hid it in Base64 and decoded it with atob(). Helme also found a fallback that runs an embedded copy when the remote script is blocked, so blocking the domain is not enough. BleepingComputer has published the indicators, and we won’t repeat them.
Once running, the payload:
- Replaced cryptocurrency wallet addresses on the page with the attacker’s.
- Captured card numbers, expiry dates and security codes typed into forms, and tried to send them out.
- Collected cookies, page URLs and screenshots.
- Loaded iPhone and iPad exploit kits. Helme identified them as Coruna and DarkSword, each aimed at different iOS versions.
The money trail is small and hard to attribute. Per Helme, the wallets in the payload had received roughly $7,000 by November 16, mostly Bitcoin (about $5,380). Some of those wallets were active before the attack. There is no public count of stolen cards, and for most merchants card data is the bigger exposure.
Sessions are the quieter risk. With a stolen session cookie, an attacker can act as a logged-in visitor without the password. The tag runs with your page’s privileges. It can rewrite content, watch form input and read the storage your origin exposes. HttpOnly cookies are out of reach. Your checkout form is not.
Who you notify, and what you do not promise
Tell customers what you can prove. If your project was one of the 188, you know two things. Your pages served a modified Optimizely file for part of November 16, and you know what that code was built to capture. You do not know how many visitors got it. Nobody does.
Pass on the consumer advice. Customers should check that day’s crypto transfers against the intended address, watch card statements and call the issuer about anything unfamiliar. They should also update iOS and iPadOS and keep an eye on any sensitive account that was logged in while they browsed.
Then bring in payments and legal. PCI DSS v4.0 requirements 6.4.3 and 11.6.1 have been mandatory since March 31, 2025. They require merchants to authorize payment-page scripts and detect tampering as the browser receives them. A vendor-side edit does not move that duty off your desk.
Under GDPR Article 33, a controller has 72 hours, where feasible, to notify its supervisory authority of a breach likely to put people’s rights at risk. If Optimizely did not contact you, confirm your file anyway. Don’t promise safety you haven’t checked.
How do you keep the next tag from rewriting checkout?
Decide which tags may run on payment pages at all. The script reached Zoom, CNN, GitLab and Salesforce visitors because those sites told browsers to trust Optimizely’s origin. Any tag allowed on a payment page can skim cards, and your own code never has to change.
- Keep experiment tags off checkout. Most tests run on landing and product pages. Payment and login pages rarely need them.
- Tighten Content Security Policy. A strict connect-src limits where page scripts can send data. That can blunt exfiltration even if the embedded fallback runs.
- Monitor what the browser receives. Subresource Integrity rarely fits experimentation assets, because they change constantly. Change detection fits, and it is what PCI 11.6.1 asks for.
On landing and product pages, keeping the Optimizely tag is reasonable once you have verified your project. Either way, check November 16 orders, support tickets about odd wallet addresses and the chargebacks that arrive over the coming weeks.
A PCI DSS penetration test shows whether your payment-page controls hold. Supply chain security testing covers the vendor code you ship but never wrote.
FAQ
Was Optimizely hacked in November 2026?
Yes. Optimizely confirmed that on November 16 an unauthorized party exploited a vulnerability in its Experimentation account signup flow and added malicious JavaScript to some customer projects. The company says it has fixed the flaw and removed the code.
Did the Optimizely malware steal credit card numbers?
The payload was built to capture card numbers, expiry dates and security codes typed into pages and send them to the attacker. There is no public figure for how many cards were taken.
How do I check if my Optimizely project was infected?
Look for Optimizely's customer notice and review your project's change history for November 16. Save the exact script your pages served before you purge caches. Then check DNS, proxy and CSP logs for the indicators BleepingComputer published.
Is the Optimizely tag safe to keep on my site?
Optimizely says the vulnerable path is blocked and the injected code removed. Keeping the tag on landing and product pages is reasonable once you have verified your project. Keep experiment tags off checkout and login pages.
Your Optimizely project may have been on the list, or you may not be able to tell yet. Either way, our data breach response team can preserve the evidence, work out what visitors were exposed to and help you word the customer notice.
Tell us what you're running
Scoping is free. We reply within one business day, and under 30 minutes for active incidents.
