Blog

Customer Advisory: SecurityScorecard Guidance Following the Optimizely Supply Chain Incident

Customer Advisory: SecurityScorecard Guidance Following the Optimizely Supply Chain Incident
What happened in the November 16 Optimizely supply chain attack, how SecurityScorecard responded, and steps your organization can take to reduce risk. Version 1.1, current as of November 17, 2026. The investigation is ongoing and we may update this statement.

Executive Summary

SecurityScorecard’s websites, platform, and products do not use Optimizely and did not serve the malicious code delivered through Optimizely snippets on November 16. At the present time, we have found no evidence that SecurityScorecard customer data was involved. We have verified our own websites and code, are completing a review of our devices and vendors, and are sharing what we have found so far so customers can do the same.

Any exposure for your organization would run through your vendors’ websites and the people and devices that visited them during the incident window, not through SecurityScorecard.

What Happened

Optimizely has confirmed that an unauthorized party exploited a vulnerability in Optimizely Web Experimentation, and inserted malicious JavaScript into customer projects. The security firm Report URI, which found the attack, published a technical analysis. It examined 1,611 snippets and found 188 infected, belonging to 178 organizations. Because it could only check the snippets it found, those figures are likely a minimum. The security firm Cynet observed the malicious code on affected sites.

According to Report URI, from 04:21 UTC until 23:11 UTC the code could swap cryptocurrency addresses, skim card details entered on affected pages, and steal screenshots and any cookies not protected by the HttpOnly flag. On iPhones and iPads not yet updated to iOS 18.7.3 or 26.3, it also reportedly delivered exploit kits through hidden frames.

The altered scripts were reportedly served from Optimizely’s normal content network and, in some cases, from customers’ own branded domains, so affected pages looked unchanged to visitors. Optimizely states that it has fixed the vulnerability, removed the injected code, and secured affected accounts. It has not published a full root-cause analysis, and Report URI reported that one cleaned snippet was re-infected during the clean-up, so we recommend continued monitoring.

Timeline (UTC)

WhenWhat
November 13 to 16Earlier wave of similar scripts reportedly served through other third-party script providers (Hudson Rock reporting; we are still assessing it)
November 16, 04:21First Optimizely injection observed by Report URI
November 16, 13:43Card-skimming function added to the malicious code
November 16, 20:40 to 21:10Mass clean-up of affected snippets
November 16, 21:52Malicious code remained in one of the 188 snippets
November 16, 23:11All monitored snippets clean; one had been re-infected after an earlier clean
November 17, 01:02Optimizely notice: vulnerability fixed, injected code removed, accounts secured

Source for the incident timeline and list of affected organizations: Report URI, “World’s Largest Brands Infected With Malware Via Optimizely.”

Our Assessment of Impact

We reviewed four areas that could connect SecurityScorecard to this incident: our properties, our code, our people and devices, and our vendors.

  • Our properties. Our public website, the SecurityScorecard platform, our trust center, support site, and API do not load Optimizely. We verified this directly on November 17.
  • Our code. Our source code does not include the Optimizely SDK or snippet. We verified this across our indexed repositories.
  • Our people and devices. At the present time, we are reviewing network and endpoint telemetry for November 13 through 16 against the published indicators.
  • Our vendors. Of the third parties we monitor in our own supply chain, 296 appear to load Optimizely on at least one website, and 33 appear on Report URI’s list of infected snippets. Of the 33, 24 match the exact domain Report URI named and 9 match a sibling domain of the same organization. The count of 296 combines different types of evidence, and we are refining it. This may affect people who visited those pages during the window. We have no indication at this time that data held on our behalf was affected. At the present time, we believe that the list includes sign-in pages at some organizations, so we are confirming with each vendor whether any page carried the snippet.
  • What our sensor recorded during the window. Our application-security evidence records the script tags on pages it visits. On the affected vendors whose scorecards we could read, it recorded Optimizely project scripts on named pages with timestamps inside the November 16 window, including at least one sign-in page. Those records indicate the dependency was live on those pages while the malicious code was being served. They do not show which version of the script each visitor received.

Our Response and Verification Steps

We verified our own environment, checked affected vendors, and reviewed the controls around third-party scripts. We are re-checking daily while the investigation continues.

  1. Verified our own properties and code base against the snippet and the published indicators.
  2. Checked the affected vendors’ current snippets. Of the 33, 13 load the snippet directly from Optimizely’s content network. We retrieved each of those and confirmed that none carries the published indicators; all were rebuilt during or after Optimizely’s clean-up on November 16. At the present time, we are confirming the remaining vendors directly with them. We are re-checking daily while the investigation continues. We have since extended this check to every Optimizely project script our sensor recorded on the affected vendors: 16 projects which, together with the 13 above, make 24 scripts in all. None carried the published indicators as of 23:30 UTC on November 17. Several were rebuilt again on November 17, which is normal experiment activity and not an indicator on its own.
  3. Reviewed the third-party scripts on our own websites and the controls around them.
  4. Compared what our different discovery methods found for the affected vendors. No single method identified every one, so we treat discovery tools as a way to shorten the list of vendors to ask, not as a way to clear a vendor.

Recommended Steps For Your Organization

Even if your own websites do not use Optimizely, your vendors may. The following steps help you find where the incident may reach your organization and reduce the risk to your users and data.

  • Use the third-party connections view in the SecurityScorecard platform (Automatic Vendor Detection) to see which of your vendors load Optimizely, and compare that set with the list of affected organizations. A vendor missing from the list is not confirmed clean. Record the hostname and date for each vendor. Treat an observed browser request or a detected library on a named host as a dependency, and treat weaker signals, such as a job posting, as a lead to ask about. Match your candidates to Report URI’s list by exact hostname first and by registrable domain second.
  • Look for the script itself in your scorecards. In the Application Security factor of your own scorecard, and of any vendor scorecard you can open, review the Unsafe Implementation of Subresource Integrity issues and look for script tags that point to an optimizely.com host. Each record names the page, the script host, the Optimizely project ID, and when our sensor saw it, which is the exact project to raise with the vendor. The record does not show which version of the script a browser received, and our sensor records a sample of the scripts on a page, so a vendor with no such record is not cleared.
  • Widen the candidate list with care. Internet-scan product fingerprints, such as the Driftnet Optimizely tag, can place a dated observation on an exact hostname. Bound the query by date and read results per hostname, because the same query can return hosts that only share a certificate or content network with the vendor. Do not clear a vendor because a tag is missing.
  • Ask affected vendors whether any authenticated pages carried the snippet, whether they rotated their Optimizely credentials, and whether they invalidated user sessions. Ask specifically about payment and sign-in pages, and ask for evidence: the preserved snippet versions with hashes and fetch times from before and after November 16, the Optimizely project change history, and confirmation that API tokens and integrations were rotated. If a vendor was cleaned and then re-infected, ask how.
  • If your staff were signed in to an affected site on November 16, sign out of all sessions and rotate any session tokens or secrets stored in cookies.
  • Block the indicator domains published in Report URI’s analysis. Search DNS, proxy, and endpoint logs for November 13 through 16 for the indicators listed below. Treat any device that reached one of the exploit-kit hosts as potentially compromised.
  • Update iPhones and iPads to iOS 18.7.3, 26.3, or later, starting with anyone who browsed on an affected site on November 16.
  • Review card activity and cryptocurrency transfers made on affected sites on November 16.
  • Reduce the risk of the next silent change. Inventory every third-party script on your own sites and its business purpose, including scripts injected by tag managers. Run a Content Security Policy in report-only mode to learn which scripts your pages really depend on, then enforce script and connection allow-lists with application owners. Add Subresource Integrity where scripts are static. It cannot be applied to the Optimizely snippet itself, whose content changes each time an experiment is published, so for that script the controls are the allow-list, change alerting on the project, and review of who can edit Project JavaScript.

Indicators (published by Report URI)

Shown with brackets so they cannot be clicked. Remove the brackets before use.

Domains: 1688360[.]cc, webshell[.]lol, linkforge[.]cc, 5vobauua3ojpklx[.]icu, eil786npto9u5bq[.]icu, l7bcsnwc7nzr3uo[.]icu, je4ft3gn8vxudum[.]icu, q1ls1ho9x7wsxlu[.]icu, ios-hack[.]icu, p55abnw2pnsz0bo[.]icu, 1y69gy2mdzaltz1[.]icu, gm38duxoihbmy5o[.]icu

IP addresses: 136.113.106[.]136, 34.96.213[.]201, 34.87.36[.]58

Block the domains and their subdomains. In script content, also look for the encoded loader pattern and the inline fallback that Report URI documented, because a search for a single domain string misses both.

What We Do Not Yet Know

Report URI’s list is a sample of the snippets it could find, not a complete count, and snippets served from customers’ own branded domains were harder to find. No public source can tell you whether a specific visitor lost data, money, or control of a device; only your own telemetry and your vendors’ records can.

We may update this advisory if our review finds anything new. For questions, contact your SecurityScorecard account team or our security team through the trust center. Your account team can walk you through the vendor view and the comparison with Report URI’s list for your portfolio.