Optimizely Supply-Chain Attack Hits 178 Sites With Card Skimmer

Attackers abused a flaw in Optimizely's signup flow to push a card skimmer, crypto clipper and iOS exploit kit onto 178 sites, including CNN, BBC and Zoom.
Table of Contents
    Add a header to begin generating the table of contents

    Attackers abused a flaw in Optimizely’s Web Experimentation platform to plant malicious JavaScript on at least 178 customer sites, exposing visitors to card skimming, cryptocurrency theft and an iOS exploit kit. Optimizely says the flaw is now fixed, but the company has not said how many projects were altered or whether the intruder still has access.

    How Attackers Turned Optimizely’s Signup Flow Into a Delivery Channel

    According to the researchers tracking the campaign, the attackers exploited a vulnerability in the account signup flow of Optimizely’s Experimentation product. That access let them add malicious JavaScript to some customer projects. Because customer websites load their experimentation scripts directly from Optimizely, every visitor to an affected page received the attacker’s code without any action by the site owner.

    Report URI counted at least 178 affected organizations. Researcher Scott Helme and the security firm Hudson Rock separately found evidence that malicious code was being served to site visitors. Named affected or exposed brands include CNN, Fox News, BBC, Zoom, Salesforce, GitLab, HSBC, IKEA and Smartsheet.

    What the Optimizely Injected Script Does to Visitors

    The payload bundles several functions. It swaps cryptocurrency wallet addresses for ones controlled by the attacker, so a user pasting a wallet address into a payment form would send funds to the wrong destination. It also skims card numbers typed into web forms, steals cookies and captures screenshots of the page.

    Together those capabilities cover payment fraud, account takeover and surveillance in a single script. The cookie theft in particular could give the attacker session access to accounts on the affected sites, though no source has confirmed how many sessions were captured.

    The iOS Exploit Kit Second Stage

    On iPhones and iPads, the script loads a second stage that researchers identified as an iOS exploit kit. Report URI said it observed no confirmed iOS device takeovers. The presence of an exploit kit nonetheless indicates the operators wanted more than card data from some visitors. A separate campaign reported the same week involved attacks on older iPhones through hijacked advertising tags on Dutch websites, though no source ties the two incidents together.

    Optimizely’s Incident Notice Leaves Key Questions Open

    Optimizely published an incident notice at 01:02 UTC on 17 November stating that the flaw had been fixed and the malicious code removed. The company said it secured affected accounts and added protections against unauthorized code delivery. The notice did not describe how the vulnerability worked, how many customer projects were modified or whether the attacker retains access.

    That last question carries weight because of what happened at Smartsheet. Its Optimizely-served script was cleaned once and then reinfected with a hidden version of the malicious code, which suggests the attacker still held access after the first cleanup.

    Customer Reactions at Salesforce and Equifax

    Salesforce said it disabled unauthorized changes to Optimizely scripts as a precaution. Equifax said it investigated, stopped using Optimizely and found no evidence that it had been compromised. The two responses show customers taking different routes: restricting the platform in one case, abandoning it in the other.

    Why Optimizely-Style Third-Party Scripts Reach So Many Sites

    The incident follows a long pattern in web security. Sites that load scripts from an outside provider extend their trust boundary to that provider, so one compromised vendor account or product flaw can reach every customer at once. Skimming operations have exploited this arrangement for years.

    What distinguishes this case is the reach of the victim list. Brands that handle payments, news traffic and enterprise sign-ins all loaded the same experimentation scripts. The number of card numbers actually exposed is unknown, and Optimizely has not published a count of altered projects, so the full scope of the theft remains unconfirmed.

    Security teams at affected sites face a verification problem. Because the malicious code was served from a trusted vendor domain, standard allow-lists would not have flagged it. Detection depended on outside researchers examining what the scripts did in the browser, which is how Report URI, Helme and Hudson Rock identified the activity.

    The reinfection at Smartsheet is the most concrete evidence that the incident was not a one-time injection. Until Optimizely publishes a technical account, customers have only the vendor’s statement that the flaw is closed.

    Related Posts