Kiberbiztonság
Optimizely attack delivers malware to major websites
Websites belonging to Zoom, the BBC and CNN were affected by an attack linked to Optimizely, according to reports. The incident shows how a third-party web script can become a tool for attempted data theft.
2026-11-17 · 5 min read
Malicious code may have reached the websites of Zoom, the BBC and CNN through a shared third-party service, the Optimizely platform, according to a report by mlq.ai. A particularly important detail is that the attack did not require the affected companies' own servers to be breached: according to SecurityWeek's reporting, the attackers reached visitors through third-party code embedded in the websites.
According to Report URI's analysis, the incident affected 178 organisations, while StepSecurity identified 186 websites. This is not simply a corporate IT problem: it shows how data handled in a browser can be put at risk even when someone visits a well-known brand's website.
The attackers gained access through the websites' shared supplier
Optimizely Web is a platform used to compare website variations through a process known as A/B testing. Such tools incorporate JavaScript code from an external provider into the way a website works. For visitors, this is largely an invisible background process, but the browser can execute the program it loads.
According to Report URI's account, the attackers inserted malicious loader code into Optimizely's Project JavaScript field. This is a piece of code designed to download and run additional code. The researchers said the infected scripts were distributed through Optimizely's content delivery infrastructure.
This is the essence of a supply chain attack: rather than attacking each target separately, attackers reach multiple organisations through a shared service. According to SecurityWeek's report, this attack likewise did not involve breaching the affected companies' own infrastructure. The distinction matters, but it does not in itself eliminate the risk to visitors.
An encrypted connection does not prove downloaded code is harmless
According to ShiftDelete's summary, the malicious code arrived over a connection with a valid security certificate. From the browser's perspective, the source was therefore not necessarily an unknown, obviously suspicious server, but a provider already used by the website.
Encryption and the trustworthiness of program code are two separate questions. A secure connection protects data in transit; it does not check what the downloaded program does. It would therefore be misleading to conclude, solely from the use of HTTPS, that every component of a page is harmless.
The attack targeted cryptocurrency transfers as well as card details
StepSecurity named the incident OPTIMJACK and described components designed to replace cryptocurrency addresses and steal payment card details. Address-replacing code, known as a clipper, aims to substitute an address copied by the user with one belonging to the attacker. This means the wallet itself does not necessarily need to be breached: the attacker is trying to change the recipient of the transfer.
Report URI's analysis also reported functionality capable of stealing cookies, taking screenshots and distributing attack tools targeting older iOS systems. These represent different avenues of attack, so the incident should not be treated solely as cryptocurrency fraud. However, the presence of a capability does not prove that it was successfully used against every visitor.
Opening an affected website, executing malicious code and confirmed data theft are not the same thing. A list of affected websites does not, on its own, establish that every visitor's payment card details or other personal data fell into unauthorised hands. The distinction between risk and actual consequences must be maintained throughout any assessment of the incident.
The figures of 178 organisations and 186 websites measure different things
Report URI reported 178 affected organisations, while StepSecurity linked 186 websites and 188 compromised script files to the incident. These figures refer to different things: an organisation, a website and a program file are not interchangeable categories. It would therefore be incorrect to automatically present these totals as conflicting counts of affected companies.
The scale of the attack is not the same as the number of victims either. Identifying an infected file or an affected website tells us about the spread of the attack, not how many people lost money. Conflating the two questions would suggest greater certainty about the consequences than the evidence supports.
Wallet transaction volume is not the same as victims' losses
According to Bitquery's blockchain analysis, transfers linked to victims totalled $2,260 across 19 separate transactions on five blockchains. The company concluded that some of the fund movements associated with the attackers were transfers between their own wallets, rather than additional victim payments. This is an important distinction when estimating cryptocurrency losses.
The amount identified by Bitquery therefore cannot automatically be treated as the final total loss from the entire incident. Transfers traced on the blockchain provide a different kind of evidence from an examination of software designed to steal payment card details. The main lesson from the analysis is that the total transaction volume of attacker-controlled wallets should not simply be added up and labelled stolen money.
Checking third-party code matters alongside fixing the flaw
According to SecurityWeek's report, Optimizely confirmed the registration flaw exploited in the attack and released a fix. Based on mlq.ai's reporting, Salesforce launched an investigation into unauthorised script modifications and disabled the affected scripts as a precaution, while Equifax stopped using the platform. Alongside addressing the provider's vulnerability, these steps also highlight the importance of checks by customers.
For web professionals, a useful starting point is to review which third-party scripts run on a site, what they do and who can modify them. It is worth separately examining whether a service is genuinely needed on every page where it currently loads. This is not just a privacy issue, but also a matter of access management and operations.
According to daily.dev's summary, a properly configured Content Security Policy, or CSP, could have restricted some stages of the attack chain. This set of rules provided to the browser can, among other things, control which sources are permitted for scripts and network connections. It is not a universal guarantee: the protection it provides depends on settings tailored to the particular website and its actual dependencies.
The most important lesson from the Optimizely-related incident is that trust in a website extends to the service providers it uses behind the scenes. Visitors should follow official notices from the affected providers, while website operators should treat third-party code as part of their own security responsibilities. A familiar brand name alone is no substitute for checking the entire technical chain.
Sources used
- 1.World's largest brands infected with malware via Optimizelyblog.report-uri.comverified
- 2.OPTIMJACK: Optimizely Hosted Scripts Crypto Clipper and Card Skimmerstepsecurity.ioverified
- 3.Optimizely Hack On-Chain Investigationbitquery.ioverified
- 4.Optimizely Vulnerability Exploited to Serve Malware on Major Websitessecurityweek.comverified
- 5.Popular websites subverted in bid to steal crypto and target iPhone userslive.euronext.comverified
- 6.Optimizely Malware Supply Chain Attackshiftdelete.netverified
- 7.Optimizely says hackers used a signup flaw to plant codemlq.aiverified
These sources were used during our editorial fact check.