Optimizely Signup Flaw Used to Inject Malicious Scripts
An attacker exploited Optimizely’s Experimentation account signup process to insert malicious JavaScript into customer projects, the company said. In its November 17, 2026 incident notice, Optimizely confirmed that it had corrected the vulnerability and removed the injected code. The company also said it had secured affected user accounts and introduced protections against unauthorized code delivery.
The service distributes website experiments using JavaScript files delivered from Optimizely’s content delivery network, according to Beazley Security. Those files can include code that customers add to their projects. Beazley’s explanation describes how malicious content inserted there can reach website visitors without any alteration to code on the website’s own server.
According to findings from security researcher Scott Helme and his team, Optimizely Web Experimentation delivered malicious JavaScript to visitors of 178 organizations on November 16, 2026. Their examination covered 1,611 Optimizely scripts, of which 188 contained malicious code. Those figures describe organizations and script files, rather than a count of affected visitors.
Scans by researchers and BleepingComputer found affected scripts on websites including GitLab, Salesforce, Fox News, Zoom, BBC and CNN. The findings established that the malicious code loaded, or that an attempt was made to load it. They did not establish that the code executed in every visitor’s browser.
Beazley Security’s code analysis described several capabilities: substituting cryptocurrency addresses displayed on a page, capturing card information entered into forms belonging to the website, and sending cookies and screenshots of pages to infrastructure controlled by the attacker. These were capabilities identified in the script, not confirmation that each function ran or that information was taken from every visitor.
The location of payment fields limited what the script could access, according to Beazley Security. Card fields contained within a payment provider’s own iframe were inaccessible to it. Details typed into forms hosted by the affected website, by contrast, were more exposed to collection by the malicious script.
Beazley Security also identified delivery of additional stages to iPhone and iPad users, linking those stages to the Coruna and DarkSword iOS exploit kits. That finding concerned delivery, rather than an established device takeover. According to The Next Web’s reporting, researchers had not documented the successful compromise of a real iPhone.
* This article is based on publicly available information at the time of writing.