What happened: Researchers identified altered Optimizely scripts associated with 178 organizations in the samples they checked. Optimizely says it fixed the vulnerable path and removed the injected code from affected projects.
What to do: If you visited an affected page, update your device and check any relevant site notifications. If you entered card details or made a crypto transfer, use the steps below. Exposure does not, by itself, prove theft or device compromise.
The Optimizely Attack: How Malicious Code Reached CNN, Zoom and Peacock
Malicious code reached pages associated with CNN, Zoom, Peacock and other major names through a website tool they trusted. The incident exposes a difficult truth: a familiar web address can be only one part of the security story.

Jump to a section
Imagine checking the address, seeing a familiar name and opening a website you have visited for years. Everything appears normal. Behind the page, however, your browser quietly loads code supplied by another company. That code has been altered.
That is the problem at the center of the Optimizely incident. On November 16, researchers found malicious JavaScript being delivered through the company's website experimentation tools. JavaScript is code that runs in a visitor's browser; sites use it for everything from interactive menus to testing different versions of a page. In this case, researchers say the injected code was designed to interfere with cryptocurrency transfers, capture payment details and deliver additional code to some iPhone and iPad visitors.
The distinction matters. A malicious script on a page is a serious finding, but it does not mean every visitor was infected, every payment was stolen or the internal systems of every named company were breached. The number of people affected and the full financial impact remain unknown. We checked the reporting on November 17, 2026.
How one supplier reached many familiar names
Optimizely sells tools that help organizations test and personalize their websites. A publisher, retailer or software company can include an Optimizely script on a page to compare different versions and see how visitors respond. That convenience also means the visitor's browser receives code from a supplier, not only from the organization whose name appears at the top of the page.
Scott Helme of Report URI reported finding 188 altered snippets associated with 178 organizations among 1,611 Optimizely scripts his team examined. The list included pages associated with the BBC, CNN, Zoom and Peacock. BleepingComputer also reported malicious code loading or attempting to load on a Fox News page. The presence of the code on a page does not establish that every part of the site or its apps was affected. Those figures describe the research sample, not a count of all affected sites or harmed visitors.
Cynet separately reported seeing suspicious browser connections and captured the insertion code while browsing CNN and Zoom. Its investigation traced the observed delivery chain to an Optimizely-hosted resource; the researchers cautioned that this did not prove every visitor was infected or that each site's own systems had been breached.
Optimizely told BleepingComputer that an unauthorized party exploited a vulnerability in the Experimentation account signup flow to add malicious JavaScript to some customer projects. The company said it blocked the vulnerable path, deployed a fix, removed injected code from affected projects, ended sessions and revoked tokens for affected accounts, and notified affected customers. It said its investigation continues.
In an update to his investigation, Helme said that by 23:11 UTC on November 16 all the infected Optimizely files his team had observed had been cleaned and remained clean as of his update. That is a finding about the files researchers tracked, not independent proof that every possible exposure or consequence has been identified. Earlier that day, their first observed injection was at 04:21 UTC; neither timestamp proves the full exposure window for every page.
These reports describe a supply-chain attack: the attacker reaches potential victims through a component another organization relies on. That is different from proving that an attacker broke into the news site's own systems, or that every page on its domain delivered the code.
What the code was designed to do
Researchers described several capabilities in the code they examined. A capability is not the same as proof it succeeded against a particular person.
- Change a crypto destination. The code looked for cryptocurrency wallet addresses and could replace them with attacker-controlled addresses. A transfer sent to the wrong address can be difficult or impossible to reverse.
- Collect payment information. The analyzed script searched for card details entered into webpages. Whether a particular payment was exposed depends on the page and how its payment form was built; a separate payment provider's cross-origin frame, for example, has browser isolation that a script on the surrounding page does not normally bypass.
- Deliver further code to some Apple devices. Researchers observed device-selective delivery aimed at iPhone and iPad browsers. Report URI identified exploit kits in the chain, while Cynet said its own evidence did not demonstrate a successful compromise of a physical device. An attempted delivery is not a confirmed infection.
The researchers also described code intended to collect information from the page, including accessible cookies. That does not establish that every login session or account was taken over. Browser protections such as HttpOnly cookies limit what page scripts can read.
Researchers identified transactions involving addresses embedded in the code, but some addresses had prior activity. Report URI considers cryptocurrency theft verified, but other researchers note that some addresses were used before this attack and transactions alone do not identify their senders. Those records do not establish a reliable total stolen in this incident. We do not know how many people lost money, had card details captured or had a device compromised.
The security decision the visitor never got to make
Most security advice asks people to look for the wrong thing: a misspelled domain, an unexpected attachment or a message from a stranger. That advice remains useful. But in this incident, the visitor could have typed a legitimate address and still encountered an altered third-party component.
Here is the practical implication: checking a web address helps you avoid impostors; it does not audit every supplier's code. HTTPS helps protect the connection in transit, but it cannot tell a browser that an authorized source is serving unwanted code. Neither fact makes those protections useless. It explains why website operators must also monitor what their pages deliver.
There is a useful parallel with the FaceStealer fake-meeting campaign: both exploit trust that feels reasonable to the person on the receiving end. The mechanisms are different. FaceStealer used familiar accounts and video to persuade targets to install malware; the Optimizely incident involved code loaded by websites. There is no evidence in these reports that the two campaigns share an operator.
What to do if you visited an affected website
Start with what you actually did and when. A company's name appearing in a list does not establish that the specific page you opened delivered the script, or that you were harmed. Check the website operator's own security notices and Optimizely's status page for updates. Avoid contacting numbers or opening links in unexpected messages claiming to help with the incident.
- If you only browsed: Install available operating system and browser security updates from the device's normal settings or the vendor's official site. This is sensible especially for iPhone and iPad users, given the attempted device-specific delivery. Do not assume an infection just because you visited, but do not assume that a lack of interaction proves no exploit could have been attempted.
- If you entered card details on a potentially affected page: Review your statements and contact the card issuer using the number on your card or its official app if you are concerned, if the merchant confirms exposure, or if a charge is unfamiliar. Ask whether the card should be replaced. The U.S. Consumer Financial Protection Bureau recommends promptly reporting suspicious transactions; local procedures and protections vary.
- If you sent cryptocurrency: Compare the completed transaction's destination with the recipient's independently verified address, using a trusted device or a previously verified record. If they differ, contact your exchange or wallet provider through an established channel immediately. Never share a recovery phrase or private key with someone offering to recover the funds.
- If a device shows signs of compromise or you receive a credible warning: Stop using it for sensitive transactions and contact your organization's IT team or a qualified security professional. Updates reduce exposure to known vulnerabilities but do not prove an existing compromise has been removed. Our guide explains how to check whether a phone may be hacked.
Do not rush to change every password solely because a brand appears on a list. If a provider confirms that a particular account or credential was exposed, follow its instructions from a trusted device, review active sessions and recovery options, and use our account recovery guide if you lose access.
What website owners should investigate
If your organization uses Optimizely, inventory the pages and environments where its scripts run. Contact the vendor through an established support channel, review project configuration and access history, and preserve relevant logs and script copies before making changes where practical. If an affected integration must be disabled, coordinate the decision with your incident-response and business teams.
Check what was actually delivered to visitors, not just what a dashboard shows now. Look for unexpected changes in third-party resources, outbound connections and affected page variants. Determine whether sensitive forms were present and how they were isolated. Review affected sessions, tokens and access with the vendor, then communicate with customers and regulators as the findings and applicable obligations require.
Longer term, treat third-party scripts as software with a business owner, an access review and an expiration plan. Reduce unnecessary scripts on sensitive pages; monitor unexpected changes; and use browser protections such as a carefully designed Content Security Policy where appropriate. These controls are layers, not a guarantee: a trusted script can change, policies have exceptions, and some versions of the payload had an embedded fallback.
The real lesson: trust needs boundaries
Visitors should not have to inspect source code before reading the news or paying a bill. The organizations asking for that trust need to understand which suppliers can run code in a customer's browser and what happens if one of those suppliers is abused.
The unanswered questions are substantial: the full exposure window, successful device compromises and actual losses still need evidence. What the incident already illustrates is clear enough. A website's security extends beyond its own logo, and a supplier's small piece of code can have a very large audience.