An attacker introduced malicious JavaScript into Optimizely customer projects and distributed it through pages on CNN, BBC, Zoom, Salesforce, and GitLab, among others. The code could change cryptocurrency addresses, capture card details, and load exploit tools for iPhone and iPad; Optimizely said it blocked the vulnerable route and removed the injected code.
A change to customer projects, not their pages
Optimizely confirmed to BleepingComputer that the attacker exploited a vulnerability in the Experimentation account registration flow. With that access, they added malicious JavaScript to some customer projects. The company said it blocked the vulnerable route, deployed a fix, and removed the code from affected projects; it also closed active sessions, revoked tokens, and reset access for affected accounts.
Optimizely said it had notified affected customers and would post updates on its status page. On November 17, that page still listed the Web Experimentation security incident as unresolved, alongside a partial outage affecting Recommendations tracking and reporting.
The platform is used to test and personalize websites. Customers load a small script, or snippet, that includes Optimizely code, customer experiments, and, optionally, Project JavaScript: custom code that runs on the pages. Report URI explained that the attacker modified this field, while the rest of each file remained intact.
As a result, the sites continued to receive the file from Optimizely's domain and with its certificate. Some customers serve the snippet from their own domains, but they were also exposed: the content still came from the same project. The incident does not, by itself, mean that the internal systems of each affected brand were breached.
188 altered scripts and a potentially larger scope
Scott Helme, a security researcher and author of the Report URI investigation, examined 1,611 Optimizely snippets and found malicious code in 188, associated with 178 organizations. Report URI noted that the actual figure could be higher because it could only review the scripts it managed to locate in time.
Identified organizations include Zoom, CNN, BBC, Fox News, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger, and Peacock. Report URI's sector breakdown includes 32 financial organizations, 23 technology and software-as-a-service organizations, 20 retail and fashion organizations, and 10 news and media organizations. Nonprofits, healthcare companies, airlines, insurers, and telecommunications services also appear on the list.
BleepingComputer verified through urlscan.io captures that the malicious script was loaded or attempted to load on pages such as Zoom, KLM, GitLab, BBC, CNN, Scheels, American Girl, ManageBuilding, Smartsheet, Stamps.com, and Salesforce. Report URI's expanded list also includes Optimizely's own sites, such as optimizely.com and login.optimizely.com.
Cynet, which investigated alerts about browsers connecting to the malicious domain, captured the code on CNN and Zoom pages and also when visiting Optimizely's website. The firm said it did not have request-and-response captures for Salesforce, Fox News, and Fox Sports, although it identified them as affected based on information from the investigation. That distinction limits what it was able to verify directly.
From a wallet address to a payment form
The modified snippets loaded an additional script from 1688360[.]cc/track.js. Report URI found three variants: a direct reference to that address, another that concealed it using Base64 and the atob() function, and a complete copy of the code that could run if the remote server was blocked. Smartsheet was cleaned up and then reinfected with the encoded variant, suggesting the attacker still had access at that point.
The main component searched for valid Bitcoin, Ethereum, and other network wallet addresses, and replaced them with addresses controlled by the attacker. According to Report URI, it could interfere with text, form fields, links, images, the clipboard, and outgoing requests. The code checked that an address was valid before changing it, and rechecked the page whenever it changed.
Card theft was added at 13:43 UTC on November 16, according to Report URI's timeline. The script monitored form submissions and POST requests for card numbers, expiration dates, and security codes; if it collected all three details, it tried to send them to 1688360[.]cc/card.php. It could not read card details entered into fields hosted in third-party payment iframes, such as those from Stripe or Adyen, but it could access details entered into the site's own forms.
When the code changed an address near words related to payments or wallets, it could also send the page URL, cookies accessible to JavaScript, the original replaced address, and a screenshot of the page generated with html2canvas to 1688360[.]cc/hit.php. Cookies marked HttpOnly were beyond its reach. The investigating companies could not determine how many card numbers or cookies actually reached the server.
Coruna and DarkSword on Apple devices
On iPhone and iPad, the script loaded a component called boot.js from two .icu domains, which checked the iOS version and selected additional content in a hidden iframe. Report URI identified Coruna for iOS 13 to 17 and DarkSword for iOS 18; for iOS 26, it described a separate chain that, at the time, only identified the device.
Between 17:13 and 17:57 UTC, the attacker added a Coruna iframe for visits from devices running iOS 18.6 or earlier. Affected pages reloaded every 60 seconds to retry the exploit, a behavior that could be noticeable. Comments included in the kit indicated that iOS 18.7.3 and later, as well as iOS 26.3 and later, were patched.
Delivery of these components does not prove that a physical device was compromised. The Hacker News reported that Cynet obtained the second stage using an emulated iPhone browser and described the result as conditional delivery, not a confirmed takeover. Cynet also warned that its captures and observed requests did not establish that all visitors had executed the code or that all had suffered data theft.
The wallets received about $7,040
Report URI estimated that the identified wallets had received about $7,040 by 19:25 UTC on November 16, over approximately nine hours. The largest amount was in a Bitcoin wallet: 0.08298 BTC, worth about $5,380. It also recorded 17.02 LTC, about $762; 662.9 USDT and 184 TRX, about $719; as well as funds on Ethereum and BNB Smart Chain.
The figures do not make it possible to attribute all those transactions to the attack. Report URI noted that some wallets were already active before the campaign and that funds were moved to other destinations between 13:03 and 13:05 UTC. The Hacker News cited a StepSecurity estimate of about $6,600 for the same addresses and noted that the firm could not determine which transfers came from victims. The difference between the two estimates does not establish the origin of each payment.
Report URI's timeline places the last clean capture of the BBC snippet at 02:21 UTC and the first observed injection into that same snippet at 04:21. The payload domain was registered at 05:23, and the first version of track.js appeared at 08:10. The script was rebuilt 26 times during the day as the attacker added capabilities and wallets.
Mass cleanup began at around 20:40 UTC. Report URI found 169 infected snippets at 20:56 and only one, belonging to TalkTalk, at 21:52. Optimizely said it had removed the code and blocked its delivery; its notice did not detail how the vulnerability worked or how the attacker gained write access to so many projects. The cause remained under investigation.
“That's for Optimizely to explain.”
— Scott Helme, security researcher
For administrators, The Hacker News recommends reviewing project and experiment JavaScript, checking which users and tokens can modify the account, and searching for references to 1688360, atob(, and unrecognized scripts. It also suggests reviewing historical logs, preserving samples, and assessing data exposure before deciding on containment measures. The investigation had not yet established the number of affected visitors or the full scope of access to information.
Sources: BleepingComputer, Report URI Blog, Report URI Blog, Optimizely, The Hacker News, Cynet Unified, AI-Powered Security Platform for MSPs & SMEs