Enterprise Cybersecurity IntelligenceWednesday

An enterprise cybersecurity intelligence company.For security and technology leaders.

Security.io Intelligence

What changed, why it matters,
and how it evolved.

Supply Chain · Lead decision brief

Optimizely’s incident report turns a vendor fix into customer response work

Optimizely has published the root cause, timeline and preliminary infrastructure for its Web Experimentation compromise.

Application SecuritySupply ChainIncident Response
Why this leads today

This ranks first because Optimizely’s 17 November report materially changed the decision from generic vendor containment to customer-specific incident response, publishing the exploited signup path, precise containment sequence, preliminary infrastructure and explicit investigation duties. It outranked today’s vulnerability disclosures because affected enterprises may already have served malicious code and must make payment, privacy and disclosure decisions before volatile evidence expires.

Read first

Treat the Optimizely event as a confirmed third-party script compromise requiring customer-specific incident evidence.

Act now

Assign web engineering to preserve project history, CDN, CSP, WAF, payment and fraud evidence.

Accountable owner

CISO with digital engineering, payments, fraud, privacy, legal and third-party-risk leadership

Decision horizon

Immediate: establish affected status and preserve evidence today; complete payment-risk and notification decisions within 24 hours.

AssessmentHigh confidence
Emerging riskA final forensic report changing affected-customer determinations, exposure windows, indicators, data-impact findings or the current statement that no other Optimizely products were affected.

What happened

The materially new development on 17 November 2026 was Optimizely’s customer report, which added the signup-flow root cause, customer-specific response guidance and preliminary indicators. On 16 November 2026, Optimizely says the first malicious Project JavaScript was written at 07:51 UTC, the activity was detected at 13:34 UTC, the signup-flow flaw was closed at 16:10 UTC and all injected code was removed by 22:24 UTC. The affected product was Optimizely Web Experimentation; Optimizely says Feature Experimentation, SDKs and legacy Full Stack were not affected.

Optimizely says the attacker created accounts through the signup flow, exploited a flaw to impersonate existing administrator users and edited customer Project JavaScript without stealing passwords. Optimizely identifies pay.pay-cloud[.]online as a skimmer host and pay-relay.wordpressadmin001.workers[.]dev as the exfiltration endpoint. A browser request from an affected site to pay-relay.wordpressadmin001.workers[.]dev is a strong sign that the skimmer ran. The script monitored pages, form submissions, clipboard events and document changes, targeting on-page payment forms and cryptocurrency transfers.

Report URI found malicious code in 188 snippets belonging to 178 organisations after checking 1,611 distinct Optimizely snippets. Report URI observed a loader for 1688360[.]cc/track.js, including plain, base64-obfuscated and inline fallback variants. The independent research also observed wallet replacement, card-skimming logic, readable-cookie and screenshot collection, and delivery paths for the Coruna and DarkSword exploit kits. The second-stage exploit attempt targeted iOS and iPadOS versions 18.6 and lower. Hosted payment iframes such as Stripe or Adyen were outside the script’s direct reach.

Attribution posture: The actor has not been named; Optimizely says attribution is pending. Optimizely’s preliminary research links pay-cloud[.]online to ADAPTAGENT and CUPCAKE C2 infrastructure used since at least September, but that infrastructure relationship is not a named-actor attribution. Optimizely engaged Mandiant and says the investigation remains ongoing. Customers must obtain their exact affected status and per-site exposure window from Optimizely rather than substituting the general incident window for customer-specific evidence. The cited source did not publish the precise quantity described as Unpublished impact quantities.

Why this matters now

The vendor’s remediation closes the upstream entry path but does not establish whether an individual customer’s visitors loaded the skimmer, submitted payment details, had wallet destinations replaced or received the iOS second stage. Customer response therefore cannot be closed with a vendor patch receipt. Web engineering, fraud, privacy, legal and incident-response owners need a shared exposure record tied to actual pages, browser requests and transaction flows.

The attack crossed organisational boundaries without requiring compromise of customer source repositories or web servers. Optimizely-hosted Project JavaScript executed inside trusted customer pages with the privileges those pages granted it. That invalidates assurance models that treat an approved script origin as permanently trustworthy and makes third-party script governance a production security control rather than a marketing-technology exception.

The response window is narrow because volatile browser, CSP, CDN, tag-manager and fraud telemetry may age out quickly. Sites using first-party card forms or cryptocurrency workflows have the highest consequence. Hosted cross-origin payment iframes reduced exposure, but that architectural protection must be demonstrated for every affected path rather than assumed from a standard checkout design.

The decision for security leaders

Treat vendor containment and customer closure as separate decisions. Optimizely can prove that its vulnerable path is closed; only the customer can prove which sensitive pages loaded the snippet, which browser requests left those pages and which transactions occurred during the relevant period. Assign one incident owner to reconcile those evidence streams and prevent fragmented technical, fraud and legal conclusions.

Prioritise first-party payment forms, cryptocurrency workflows and pages without enforced CSP. Do not infer safety from an unaffected corporate network, clean application repository or trusted CDN certificate: the malicious code executed in visitors’ browsers through an approved dependency. Where evidence is incomplete, record the limitation explicitly and make the resulting risk acceptance or notification decision at the appropriate authority level.

Use the incident to reset third-party JavaScript governance. Require named business ownership, sensitive-page approval, immutable change records, browser-side integrity monitoring, outbound destination controls and periodic removal of unused tags. Hosted cross-origin payment iframes and strong CSP controls should become architectural requirements for high-consequence pages rather than optional compensating measures.

Evidence of closure

  • Optimizely confirmation records the customer’s affected status and exact exposure window.
  • Preserved logs show no requests to vendor-published attacker domains from sensitive pages.
  • Project-history diffs show unauthorised JavaScript removed and legitimate code restored.
  • Payment and fraud review records an approved disposition for each exposed transaction path.、CSP enforcement testing blocks an unapproved script and exfiltration destination.

The Security.io assessment

The new report materially improves the entry-path and containment picture, but it does not collapse the customer investigation into a simple patch confirmation. Optimizely’s direct logs can identify altered projects; they cannot by themselves prove every visitor-side outcome, successful submission, wallet replacement or second-stage result. The decisive evidence remains distributed across the vendor, the customer’s browser and edge telemetry, payment systems, fraud operations and customer-specific application architecture.

Report URI’s observations and Optimizely’s preliminary findings substantially agree on the browser-delivered skimmer mechanism, while differing terminology and observation times should not be treated as contradictory. The most defensible response is to preserve both data sets, privilege customer-specific vendor records over generic lists and document where independent telemetry confirms or cannot confirm execution and outbound communication.

No count of payment cards captured, affected visitors or successfully compromised iOS devices was published. Absence of a published count is not evidence of zero impact. Conversely, observed malicious capability does not prove that every infected snippet captured data or compromised a device. Closure requires an evidence-backed disposition for each sensitive page and transaction path, with unresolved gaps escalated rather than silently converted into an unaffected conclusion.

Questions for the morning meeting

  • Can we prove whether Optimizely Web Experimentation ran on checkout, authentication or cryptocurrency-transfer pages?
  • Do retained browser, CSP, CDN and proxy logs cover each customer-specific exposure window?
  • Who owns payment, privacy and regulatory decisions if visitor exposure cannot be conclusively excluded?
  • Which other third-party scripts can execute with first-party privileges on sensitive pages?

Related intelligence

Shared decision context