FR
live

A hacker hijacks Optimizely scripts and serves a card skimmer to CNN, BBC, and Zoom

On November 17, 2026, Optimizely confirmed that a flaw in its A/B testing product let an attacker inject malicious code into scripts loaded by 178 organizations, including CNN, the BBC, and Zoom. Inventory the third-party scripts your pages run and lock down your vendors’ custom JavaScript.

A credit card payment terminal with a thin clip-on skimmer overlay covering its card slot, a single amber LED lit on the overlay.

November 17, 2026. Optimizely confirms an attacker abused a flaw in its Web Experimentation product to inject malicious code into customer projects. November 16, 2026, 04:21 UTC. The first injection, spotted in the BBC’s snippet. 178 organizations. The number of customers whose scripts served the code, per Report URI. Why it matters: nobody attacked CNN, the BBC, or Zoom directly — their A/B testing vendor was turned against them, and the code shipped from the vendor’s own CDN, under the vendor’s own certificate.

The payload: skimmer, crypto theft, and an iOS exploit kit

The payload, a file called track.js, had three jobs. The first targeted crypto wallets: it swapped valid addresses for the attacker’s wherever they appeared — in text, forms, links, the clipboard, and outgoing requests. It only replaced an address that passed its chain’s checksum and sat near a payment word (wallet, deposit, send), an analysis confirmed by StepSecurity. Copying an address offered no protection: it could already be the attacker’s by the time it was pasted into a wallet.

The second job was a card skimmer, added on November 16 at 13:43 UTC. It captured the card number, expiry date, and CVV typed into the site’s form fields and sent them to the attacker’s server. Its one limit: it could not read cards entered in a cross-origin payment iframe, such as Stripe’s hosted fields or Adyen’s.

The third job applied only to iPhones and iPads. There, the code loaded a second stage that Report URI identifies as two iOS exploit kits, Coruna and DarkSword, documented by Google researchers. Coruna targets iOS 13.0 through 17.2.1. No iPhone is reported to have actually been taken over, but the plumbing was in place.

The code also sent the visitor’s non-HttpOnly cookies and a screenshot of the page to the attacker — but only after it swapped an address near wallet or payment wording, which caps the exposure.

Project JavaScript: the door the flaw opened

The entry point is not an exotic exploit. Optimizely customers load a snippet provided by the vendor, which updates itself whenever a setting changes. On every load, that snippet runs a field of custom code called Project JavaScript. That is where the attacker wrote the loader — and it left the rest of the snippet untouched.

Three loader variants circulated. The most common loaded 1688360[.]cc/track.js directly. A second hid that address in base64 to dodge a text search. A third carried a full copy of the payload, ready to run if the domain was blocked.

The worst part is elsewhere: the snippets kept being served from cdn.optimizely.com under the vendor’s certificate. A page loading a healthy-looking snippet was in fact running the loader. Even sites serving the snippet from their own domain were affected, because the content came from the same project. Visual inspection or a domain inventory would not have caught it.

Report URI checked 1,611 snippets — mostly from its own crawl data — and found the code in 188 of them, belonging to 178 organizations. The real number is likely higher: the firm could only inspect snippets it found in time. Beyond CNN, the BBC, and Zoom, the list includes Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger, and Peacock. Optimizely’s own main site and login page carried the code too. Some researchers call the campaign OPTIMJACK.

What the vendor did not say

The incident notice posted on November 17 at 01:02 UTC amounts to one sentence: the attacker “exploited a vulnerability in Experimentation” to add malicious JavaScript to some projects. Optimizely says it fixed the flaw, removed the code, and secured affected accounts.

What is missing says a lot. The notice does not say how the flaw worked, how many projects were changed, when the fix went in, or whether the attacker still holds access. Report URI, writing before the notice appeared, summed it up: “That’s for Optimizely to explain.”

One detail suggests access was not cut immediately: Smartsheet’s script was cleaned once, then re-infected with a hidden version of the code. And the same injector, webshell[.]lol, had already been planted through other third-party scripts — Kissmetrics, StorifyMe, and Minute Media — since late the previous week, first surfacing on Turkish sites on September 17.

Why nobody saw it coming

This kind of attack is structurally hard to detect. The malicious code never touched your servers or your repositories: it flowed through the vendor’s CDN, signed by the vendor’s certificate, and left no trace in your application logs. A SIEM watching your systems has nothing to see here — the compromise plays out entirely inside your visitors’ browsers.

It is the lineage of Magecart, the collective that popularized web skimming from 2015 onward by injecting skimmers into third-party payment scripts. The novelty is not the technique: it is the scale and the target. Instead of compromising one site at a time, the attacker compromised one vendor and hit 178 organizations at once — the supply-chain multiplier. The same injector had already been planted in Kissmetrics, StorifyMe, and Minute Media scripts, which points to a rolling operation across the third-party script ecosystem rather than a one-off against a single vendor.

Detection therefore demands tools that watch what the browser actually executes, not what your servers serve. Report URI caught the campaign precisely because it monitors the scripts sites load inside their visitors’ browsers. A CSP in report-only mode, wired to a collection endpoint, gives you the same visibility: every unlisted script fires a report before it is even blocked. That telemetry is your earliest warning, because the moment a vendor silently changes the code it ships, the violation log moves first.

The verdict

If you run Optimizely — or any A/B testing tool that executes custom code — do three things right now. First, pull your projects’ Project JavaScript and audit it line by line: it is code that runs on every page, not a configuration toggle. Second, lock it down: disable custom code wherever it is not essential, and demand a clear access model from the vendor — who can write, from where, with what logging. Third, add page-side tripwires, a script-allowlist CSP and SRI (subresource integrity) on third-party scripts, so a future hijack fails at load time instead of executing.

For everyone else, treat this as a reusable template. Your attack surface includes every third-party script your pages load, and a compromised vendor ships code under its own certificate, invisible to most of your controls. Inventory those scripts, interrogate their execution rights, and treat your vendors’ custom JavaScript as production code — because it is. A hijacked update channel, like cPGuard’s backdoor of 3,000 Linux servers, differs only in the channel: here, the channel is your own homepage.

The timeline also shows how fast a single write access cascades: the first injection landed at 04:21 UTC, a card skimmer followed at 13:43 UTC, and within eighteen hours the code had reached the snippet of nearly every major brand on the list. Speed of detection is the only counterweight to speed of injection.

References

The cyber brief, every Tuesday

The flaws that matter and the patches to apply, in a ten-minute read.

No spam. One-click unsubscribe.
read next

On the same topic

UnrealIRCd 6.2.7.1 closes an exploited flaw that hands server authority to any user

CVE-2026-114713, rated 8.6, lets any user of an IRC network with linked servers inject server-to-server commands through the label tag of the labeled-response module. Added to CISA’s KEV catalog on 13 November 2026 with a 16 November deadline, it is already being exploited: apply the hot-patch without restarting, or blacklist the module.

← Back to the feed

Type at least two characters.

↑ ↓ navigate ↵ open esc dismiss