Optimizely malware hit scripts at 178 organisations, Report URI says | Magica
Report URI says Optimizely scripts carried malware across 178 organisations
Editorial Team
••📖8 min read
Malicious code in Optimizely website-testing projects could swap cryptocurrency addresses and skim payment cards. Optimizely says it has fixed the vulnerability and removed the code, but visitor exposure and losses remain unquantified.
Report URI found 188 infected Optimizely script snippets belonging to 178 organisations in a crawl of 1,611 snippets.
Optimizely says it fixed the vulnerability and removed the injected code; the number of affected visitors remains unknown.
Wallet receipts do not establish total incident losses, and isolated payment fields limited the card skimmer's reach.
Browser-security monitoring provider Report URI found malicious JavaScript in Optimizely website-testing scripts belonging to 178 organisations on 16 November 2026. The finding measures infected customer scripts, while the number of visitors who executed the malware remains unknown. Its investigation identified code capable of replacing cryptocurrency wallet addresses, stealing payment-card details and loading additional malicious code on Apple mobile devices.
Report URI associated the infected scripts with domains used by organisations including Zoom, the BBC, CNN, GitLab, HSBC and Ledger. That does not establish that every page or service operated by those organisations served the payload. The central failure was in a dependency those websites trusted: code delivered through a legitimate experimentation service could act inside visitors' browsers, a route that does not require breaching each website's own infrastructure.
Optimizely, whose Web Experimentation product lets businesses test website changes, has acknowledged the incident. In a status notice posted at 01:02 UTC on 17 November, it said an unauthorised party exploited a vulnerability in Experimentation to add malicious JavaScript to some customer projects. The company said it had fixed the flaw, removed the injected code, secured affected accounts and added protections against unauthorised code delivery.
The platform sits within a broader software business: Episerver acquired Optimizely in 2020 and adopted its name in January 2021, combining website experimentation with content and commerce capabilities. Here, the consequential feature was customers’ ability to publish code directly into pages through the experimentation service.
Report URI’s investigation screenshot shows Optimizely and track.js script references in Zoom’s page. Credit: Scott Helme / Report URI, CC BY-SA 4.0. .
The management company supporting Japan's Shonan Beauty Clinic network disclosed stolen customer passwords stored without encryption and a separate booking-service incident. Treatment-related submissions were also exposed, while the affected population and financial impact remain under investigation.
A legitimate publishing mechanism carried the payload
Optimizely Web delivers each customer's experiments through a JavaScript snippet. Its Project JavaScript field allows customers to add custom code that executes wherever the snippet is included. Report URI found the malicious addition in that field, with the rest of the snippet untouched.
The injected loader fetched an attacker-controlled script, track.js, from 1688360.cc. The infected snippet still came from Optimizely's trusted delivery domain. Report URI also found affected snippets served from customers’ own domains: moving the delivery hostname did not change the underlying compromised project content. Report URI explicitly distinguished this from CDN cache poisoning: the delivery network served the content it had been given.
An Optimizely spokesperson gave a more specific account in a reported response, saying the vulnerability was in the Experimentation account signup flow. Optimizely said it had blocked that path, ended active sessions, revoked tokens, reset access for affected accounts and notified affected customers. The available explanation does not detail how the flaw permitted changes across unrelated customer projects.
Independent checks support the finding that legitimate snippets carried the loader. A download of ten websites' Optimizely files between 20:33 and 20:36 UTC on 16 November found six infected and four clean. That small check also demonstrated that infection was not universal among customers.
Report URI's broader crawl found 188 infected snippets belonging to 178 organisations among 1,611 distinct snippets, mostly discovered through its own crawl data. Those are different units: one organisation can have multiple snippets. The crawl was not a random sample of all Optimizely customers, and the investigators said the true number could be higher.
A separate payload analysis cited Report URI chief executive and security researcher Scott Helme for a count of 186 infected sites and 188 script files. That site count differs from Report URI's published organisational tally; it is not an independent census, and neither figure measures how many visitors received or executed the malware.
The campaign's broader delivery routes also require care. An initial analysis naming it OPTIMJACK traced twelve of fifteen example page loads to Optimizely-hosted scripts; three involved an advertising wrapper, a first-party bundle or another protected script. That evidence does not establish an identical loader chain for every affected page. Report URI also linked the actor to an earlier wave delivered through Kissmetrics, StorifyMe and Minute Media scripts between 13 and 16 November. It excluded those additional sites from its Optimizely tally; the claim concerns scripts carrying the code, rather than a demonstrated common vulnerability across those providers.
The theft mechanisms had different limits
The cryptocurrency component rewrote valid wallet addresses in page content, forms and clipboard operations. A visitor could copy what appeared to be a recipient's address and paste the attacker's address into a wallet application. Blocking outbound theft requests alone would not stop that manipulation, because the substitution happens within the page.
The capabilities also changed over time. Report URI recorded 26 payload builds during the day and put the addition of card skimming at 13:43 UTC. An early visit and a later visit therefore cannot automatically be assigned the same exposure.
The card-skimming component searched form submissions and outgoing requests for a card number, expiry date and security code, then sent captured details to the attacker's server. Both Report URI and a separate payload analysis found an important boundary: the script could not read payment fields isolated in a payment provider's cross-origin iframe. Card-entry fields in the affected page’s own forms were within the script’s reach. The separate analysis also found that captured cards were stored in the visitor’s browser session storage to prevent duplicate submissions, leaving an additional local copy of the data.
Screenshot and cookie collection was tied to wallet-address swapping, rather than occurring for every visitor. Cookies marked HttpOnly were inaccessible to the script. These limits matter when assessing exposure: loading an infected snippet, executing the payload and losing data are separate events.
On iPhones and iPads, the payload loaded a further stage. Report URI identified Coruna and DarkSword exploit chains. The separate analysis confirmed the mobile loaders but said it had not retrieved or analysed those exploit files and could not independently confirm which chain ran against which versions. Neither finding supplies a count of successfully compromised devices.
Wallet receipts are not a loss estimate
Report URI put receipts into attacker wallets at about $7,040 as of 19:25 UTC on 16 November and described cryptocurrency as already stolen. The separate wallet analysis put funds received or held at roughly $6,600, mixing cumulative receipts for some chains with a current balance for another and using 16 November prices. It also found activity in some addresses before this campaign and cautioned that blockchain records do not identify which transfers came from victims.
Those figures should not be treated as a settled incident loss total. They do not measure stolen card details, cookies or mobile-device access, and Report URI said it could not see how many cards or cookies reached the collection server. None of the investigations quantified the number of exposed visitors.
The control problem has a precedent. In its findings on Ticketmaster's 2018 breach, the UK's Information Commissioner's Office said placing a third-party-hosted chatbot on a payment page allowed an attacker access to customers' financial details. The regulator fined Ticketmaster UK £1.25 million in November 2020 and said the breach potentially affected 9.4 million customers across Europe, including 1.5 million in the UK. Those are historical figures, not estimates for Optimizely. The relevant comparison is the authority granted to external code on a sensitive page and the customer organisation’s responsibility for that choice.
Report URI’s observations show the payload changing before cleanup. These milestones do not establish individual visitors’ exposure. Source: Report URI.
The next test is a customer-level exposure account
Report URI observed the BBC's first infected Optimizely capture at 04:21 UTC on 16 November. By 21:52 UTC, it found only one of the 188 identified snippets still infected. Optimizely's subsequent notice says the injected code has been removed. These observations describe successive stages of cleanup, rather than proving that earlier visitors escaped exposure. Report URI also saw a Smartsheet snippet cleaned and then reinfected with an encoded loader, evidence that removal alone had not ended the attacker’s access at that point.
For customers, a clean snippet now cannot resolve what it served during the incident. The needed evidence is a project-level timeline of unauthorised changes, the pages that included each snippet and browser or network records showing which payloads executed.
The investigations point to layered controls: remove unnecessary third-party tags from checkout, restrict which additional scripts may execute, limit destinations for outbound data and isolate payment fields. Content Security Policy (CSP) is the browser mechanism used to restrict script sources and data destinations. There is an operational tradeoff: experimentation snippets change when teams publish tests, making a permanently fixed integrity hash difficult to maintain. Report URI also sells monitoring and enforcement tools for this problem, a commercial interest relevant to its recommendations.
Policy details also change the result. Report URI found an inline payload variant that could use eval to run embedded code even when the external download was blocked. It warned that permissive evaluation settings could allow that fallback, while strict-dynamic can extend a trusted script's authority to scripts it loads. An allowlist containing Optimizely's hostname alone therefore does not establish protection against the observed variants.
Optimizely has supplied an initial account of the vulnerability and containment. A fuller explanation still needs to establish how the signup flaw crossed project boundaries, the complete affected population and the resulting visitor harm. Customers need that evidence to decide whose payment details or sessions require action, and what they should tell visitors about the incident.
N-able says an intruder reached a server supporting Mail Assure and SpamExperts Hosted Cloud. Mail delivery and filtering continued, but customers lost management access as the company revoked tokens and worked to restore services securely.
Microsoft is testing an optional Edge mode that blocks WebGL and WebGPU to reduce opportunities for browser exploitation. An early hands-on test found that restoring graphics for a trusted site also removed its broader Enhanced Security protections.
A critical authentication bypass gave a suspected China-linked espionage group access to VMware's virtualization control plane. Mandiant says the campaign stole signing keys and virtual-machine data, making compromise checks essential alongside patching.
PHP’s November 19 security releases address seven critical flaws, including unauthenticated code execution through reused backend connections. Patching takes priority, but applications that retain session upload progress must also check shared storage and serialization settings.
TfL has shut its Oyster photocard website after a breach that may have exposed around 265,000 customers’ account details, including security answers. Existing cards still work, while the attack method and restoration timetable remain undisclosed.
Around 40 Irish credit unions are implicated in a saved-payee data breach at software supplier Wellington IT. Missing logs leave the victim count unresolved, while a minister is seeking answers about the disclosure delay.
JPMorganChase says it believes attackers accessed marketing data at outside service providers, potentially exposing customer names and email addresses. The bank says services are unaffected, but has not identified the vendors or disclosed the attack date.
Apache confirms exploitation of a Roller code-execution flaw that survives disabling XML-RPC in its administration panel. Attacks on the retired Axis 1.x library also expose enterprise applications, making product-specific mitigation and compromise checks urgent.
StepSecurity’s Elephant-led Series A will fund broader protection for developer machines, package downloads and build pipelines, alongside planned defenses for AI-agent sandboxes. Its challenge is turning open-source adoption into a larger commercial platform while proving the extra controls earn their place.
Nvidia says it has found no indication its own systems were affected after The Gentlemen ransomware group listed it on a leak site. The investigation concerns a possible incident at a partner, while the gang has supplied no public proof.
Senators from both parties plan to delay the CISA nomination of Palantir security chief Michael LoSapio after his Jan. 6 denial, adding a confirmation obstacle as the agency faces staffing losses and funding decisions.
Sean Cairncross is leaving as national cyber director, with Pentagon CIO Kirsten A. Davies reportedly selected to succeed him. The transition puts Senate confirmation at the center of a cybersecurity leadership reshuffle.
Windows baseline protection combines existing PC safeguards into a verifiable pass-or-fail result. Microsoft defines the standard; developers choose whether failure means a warning, fewer features or blocked access.
F5 lists fixed releases for a critical nginx HTTP/2 flaw, while the wider NGINX2Root disclosure covers five independent nginx attack paths and a separate Linux kernel bug. Remediation depends on configuration, supplier patch levels and replacing vulnerable running processes.
Tyler Robert Buchanan received six years in U.S. prison for a phishing and SIM-swapping conspiracy that stole at least $11.6 million from individuals. He is the last of five defendants sentenced; restitution remains pending.
The new Threat Disruption Center makes TRM Labs its main technology partner following a reported $94.66 million sole-source award for Homeland Security Task Force investigations. A rival lost its court challenge, but the launch leaves the center’s share of that spending and its performance benchmarks undisclosed.
Fandango says payment details from certain October 8–11 transactions may have been exposed. State disclosures provide only a partial affected count, while the company has not explained how the checkout page was compromised.
Norway’s state-controlled oil and gas group is working to neutralise an attack on some IT systems. Employee services may be disrupted, but Equinor says operations and production remain unaffected and it has found no indications of data theft.
Tor urges relay operators to install 0.4.9.17 as a continuing DDoS attack strains its privacy network. A post-upgrade operator account describes further disruption, while Bitcoin connectivity measurements show partial recovery with important limits.