Optimizely confirmed attackers exploited a vulnerability in its Experimentation product to inject malicious JavaScript into customer projects, and researchers found 188 infected scripts tied to 178 organizations including the BBC, CNN, Zoom, Salesforce, GitLab, Equifax, HSBC, and IKEA. The injected code could swap cryptocurrency wallet addresses, skim payment card details, collect cookies and screenshots, and serve iOS exploit kits to iPhone and iPad visitors. Roughly $7,000 in cryptocurrency reached the attacker before cleanup finished. Optimizely patched the flaw and removed the code, but has not said how many customers were affected, and researchers found similar loaders on other analytics providers' scripts dating back to November 13.
What Happened
Optimizely, a widely used digital experience and testing platform, disclosed on November 17 that an unauthorized party had exploited a vulnerability in its Experimentation product and used it to inject malicious JavaScript into customer projects. The company posted updates to its status page and on social media, confirming that some Web Experimentation customers were affected, and said it had fixed the vulnerability and removed the injected code from affected projects. It also said it had taken steps to secure affected user accounts and introduced additional protections, though it has not disclosed how many customers were impacted overall. Independent researchers moved quickly to assess the scope. Report URI, a security firm, examined 1,611 Optimizely scripts in circulation and identified 188 infected files belonging to 178 distinct organizations. The list of affected organizations includes some of the world's most visible brands: the BBC, CNN, Zoom, GitLab, Salesforce, Equifax, HSBC, and IKEA among them. According to Report URI's analysis, the malicious code was inserted into an optional field within Optimizely's platform meant for a customer's own custom JavaScript, a field that executes on every single page load. Because the tampered files were served from Optimizely's own content delivery network, any website loading them treated the code as fully trusted, with no reason to suspect tampering. Report URI's timeline traces the first injection to November 16 at 04:21 UTC. By 21:52 UTC the same day, all but one of the infected scripts had been cleaned, and by 23:11 UTC the remaining file was also clear. During that roughly 18-hour window, the malicious code was capable of multiple forms of abuse: it could swap cryptocurrency wallet addresses displayed on a page for the attacker's own address, skim payment card details entered by visitors, and collect browser cookies and screenshots of the page content. Report URI estimated that approximately $7,000 in cryptocurrency reached the attacker's wallet during the incident, a modest but concrete financial outcome that confirms the attack was at least partly monetized before detection. A separate and more concerning thread involves mobile exploitation. Report URI found that iPhone and iPad visitors to affected sites were served the Coruna and DarkSword iOS exploit kits. Hudson Rock, another security firm, corroborated the connection to exploit-kit delivery infrastructure but was careful to note the limits of current evidence: it found no proof that a zero-day vulnerability was used, that any device was successfully exploited, or that any specific device data was actually collected by whatever downstream stages the exploit kits might trigger. Hudson Rock's broader investigation, which drew on findings from a researcher at a firm called Ohey, also found that Optimizely was not the only delivery channel in this campaign. Scripts belonging to other analytics and marketing providers, including Kissmetrics, StorifyMe, and Minute-ly, were observed serving a smaller loader that led to the same iOS-targeting infrastructure, with the earliest sighting dating back to November 13, three days before the Optimizely incident was publicly confirmed. It remains unclear exactly how the malicious code reached these other providers' scripts, raising the possibility of a broader and still partially unmapped campaign against the analytics and personalization software ecosystem that many websites depend on.
Why It Matters
This incident is a textbook supply-chain attack: rather than compromising 178 separate organizations' own servers, the attacker found a single point of leverage, a shared third-party service, that automatically propagated malicious code to every website using it. That structure matters because it inverts the usual security assumption that a website's own hardening determines its safety. Scott Helme of Report URI summarized the core lesson plainly: trusted JavaScript doesn't always remain trustworthy, and a website can be compromised through a third-party dependency without its own code or infrastructure ever being breached. For the organizations named, reputational exposure is real even though their own systems were never touched, since customers and visitors will associate the incident with the brand they were visiting, not the vendor whose code actually failed. Hudson Rock's warning that some of the poisoned scripts carried a full embedded version of the payload is also significant: it means simply blocking the attacker's remote domain may not fully neutralize affected sites, since a locally embedded fallback could still execute if the external file failed to load.
What to Watch
The full customer count affected by Optimizely remains undisclosed, so additional organizations beyond the 178 identified by Report URI could still surface as further scanning continues. The connection between the Optimizely injection and the separate loaders found on Kissmetrics, StorifyMe, and Minute-ly is also unresolved; if investigators confirm a common entry point or shared attacker infrastructure across these unrelated providers, it would suggest a more coordinated campaign against the broader analytics and personalization software supply chain rather than an isolated incident at a single vendor. Whether the iOS exploit kit activity ever achieved successful device exploitation, as opposed to simply attempting delivery, also remains an open and important question that could change the severity assessment considerably.
What Defenders Should Do
Security teams should inventory every third-party script currently loaded on public-facing websites, with particular attention to analytics, testing, and personalization tools that execute on every page view. Teams should apply or tighten content security policies and subresource integrity checks so that any unexpected change to a trusted vendor's script is detected or blocked automatically rather than silently trusted. Organizations that use Optimizely, Kissmetrics, StorifyMe, or Minute-ly should review their own traffic logs for the November 13 through 17 window for any signs of the described payload behavior, and should treat any cryptocurrency-related transactions conducted by site visitors during that period with added scrutiny.
The Larger Lesson
Trust in a website is increasingly borrowed from a chain of third-party vendors whose own security posture a visitor can never see. As personalization, analytics, and testing tools become standard on nearly every commercial site, the number of silent dependencies that can be weaponized grows accordingly, and this incident demonstrates that even globally recognized, well-resourced organizations can be compromised through code they never wrote and never directly control.
