

Variedy Security Alerts
Cybersecurity, Threat Intelligence
Read
178 Companies Hit: Inside the Optimizely Supply-Chain Attack
A Variedy Cybersecurity Series
Coming off a summer defined by AI agent swarms escaping labs and hacking websites across the world—including OpenAI’s Sam Altman halting the release of its latest Astra model after learning it broke through its training sandboxes to swarm and hack Hugging Face—we have faced daily reports of escalating security threats, culminating in internet PHP exploits that rank among the worst in web history. We thought it couldn’t get worse, yet November is shaping up to be one of the most serious months for cybersecurity in the history of the computer age.
With Microsoft Outlook services going dark and causing business interruptions worldwide, military and global institutions have issued stark warnings urging citizens to stock up on food and water in case of an electrical grid blackout. Cybersecurity professionals remain on high alert. The past week ended with U.S. air traffic control grounding flights nationwide, leaving millions of American travelers stranded in high-stress situations. As unanswered questions poured in over the weekend, Monday morning arrived with hundreds of websites suffering outages and major retail banks and financial institutions going down—cutting millions of Americans off from their accounts and funds.
For over 25 years, Variedy has been a leader in helping enterprises and institutions across all industries secure their IT infrastructure, cloud environments, and on-premises systems. At no time in history has it been more critical to have up-to-date information, factual reporting, and actionable solutions the moment they become available.
Security Alerts: Behind the Scenes is an ongoing series dedicated to uncovering the technical realities behind the headlines. Most of these reports are covered in real time and will be updated continuously as new details emerge. Please check the latest update timestamps on both our articles and videos.
Last Updated: November 17, 2026
If you opened the BBC website on Monday morning, November 16, 2026, your browser ran code written by a stranger. It was the real BBC site, at the real web address, with the security padlock in the corner. Nothing looked wrong.
The exact same malicious code ran on Zoom, CNN, Fox News, NPR, HSBC, Equifax, Salesforce, GitLab, IKEA, KLM, and the crypto wallet maker Ledger. Security researchers counted 178 affected organizations.
Yet most of those companies weren't actually hacked. Nobody broke into the BBC's servers, and nobody breached Zoom’s internal network. Instead, the code arrived through a small, behind-the-scenes tool almost nobody outside of marketing and web development has ever heard of—a tool those sites trusted so deeply that they loaded it first, before their own pages.
So, is the headline true? Did one cyberattack really hit 178 major organizations? Yes—though not in the way most people picture it. Here is what happened hour by hour, and what that hidden code was built to steal from you.
What "Hit" Really Means
The 178 figure comes from security researcher Scott Helme, founder of Report URI, a company that monitors the third-party scripts websites load. On Monday, Helme's team pulled 1,611 copies of a script from a single provider: Optimizely. Out of those, 188 files carried hidden malicious code, belonging to 178 distinct organizations. Report URI notes the true number is likely higher, as they could only inspect the files they discovered within the attack window.
(One fair caveat: Report URI sells monitoring tools designed to detect this exact type of attack, a point Gizmodo highlighted in its coverage. Wherever possible, we have cross-checked Report URI's figures against independent security firms.)
In this context, being "hit" means visitors to those websites were served harmful code inside a file the website explicitly asked their browser to load. It does not mean hackers infiltrated those companies' internal corporate networks.
Equifax confirmed it found no evidence of any compromise to its internal systems, though Reutersreported the company stopped using Optimizely anyway.
For IKEA, the exposure was isolated to a single home-design subdomain rather than its main retail store. Security firm Hudson Rock emphasized that an affected subdomain does not establish exposure across an organization's entire web presence.
Picture a single poison delivery dropped onto 178 front doorsteps.
How One Text Box Opened 178 Front Doors
To understand how one breach reached so many major brands at once, you have to understand what Optimizely does.
Have you ever noticed an online store looking slightly different from one visit to the next—a green "Buy" button one day, an orange one the next? That is A/B testing. Companies show half their visitors one version and half another, keeping whichever converts better. Optimizely is one of the largest platforms in the world for running these tests, boasting more than two million experiments run on its service.
Here is how it works under the hood:
The Snippet: A website embeds a single line of code on every page that fetches a small JavaScript file—called a "snippet"—directly from Optimizely’s servers.
Instant Edits: When a marketing team updates a test in Optimizely's dashboard, that snippet updates automatically without the website's own software engineers ever touching a line of code.
Top-Priority Loading: To swap a button before your eyes catch the old version (a visual glitch known as "flicker"), Optimizely instructs customers to place the snippet as high up in the page's
<head>tag as possible so it runs before almost anything else.
Crucially, Optimizely includes an advanced setting called Project JavaScript (projectJS)—essentially a custom text box in the dashboard that lets customers run arbitrary code before the main experimentation snippet executes. According to Report URI, whatever code sits in that text box runs on every single page. That text box is where the attack lived.
Hour-by-Hour: How Monday Unfolded
All times below are in UK time (UTC/GMT).
02:21 AM: An internet scanning service captures the BBC’s Optimizely snippet. It is completely clean.
04:21 AM: A new scan catches the BBC's file infected. A few lines of code had been injected into the Project JavaScript box, instructing every visitor's browser to fetch an external script from an attacker-controlled domain. The rest of the file was untouched, still served from Optimizely’s official content delivery network (CDN) with Optimizely's valid security certificate.
05:23 AM: After briefly pulling from an older server, the attacker registers a brand-new domain (
1688360.cc) to host the main payload for the rest of the day.08:10 AM: The first version of the main malicious script goes live, initially built to steal cryptocurrency. By 08:28 AM, it is actively running on
bbc.com.11:20 AM: Zoom’s snippet is modified, alongside Salesforce, CNN, Fox News, Air France-KLM, GitLab, Smartsheet, and Stamps.com. Throughout the day, Report URI tracked 26 separate builds of the main script—the attacker kept adding new features and wallet addresses without ever removing old ones.
01:43 PM: The attacker adds a credit card skimmer module to the script.
05:00 PM – 06:00 PM: The script grows a third arm specifically targeting iPhones and iPads.
07:24 PM: Report URI completes a wide sweep, finding 186 infected files out of 1,611 scanned. At 08:10 PM, Scott Helme breaks the news on X, warning of a staggering third-party JavaScript breach impacting hundreds of high-value sites—a post quickly viewed millions of times.
08:40 PM – 11:11 PM: Cleanup moves rapidly. Ledger's file is scrubbed at 08:40 PM, Smartsheet's at 08:56 PM, and the BBC's at 09:07 PM. By 09:52 PM, only one infected snippet remained (at UK broadband provider TalkTalk), and by 11:11 PM, every tracked Optimizely file was clean.
Step Security estimates most affected sites served the poisoned script for roughly 16 hours. However, the cleanup wasn't smooth everywhere: after Smartsheet's file was cleaned, the attacker re-infected it using obfuscated text to hide the script's web address—proving the intruder still had active access during the remediation window. Furthermore, on the BBC and one HSBC snippet, the attacker packed a full 24-kilobyte backup copy of the malware directly inside the Optimizely file. If a visitor's network blocked the external domain or took more than 2.5 seconds to load it, the built-in copy fired automatically.
The Malware's Four Jobs
Report URI confirmed that every visitor received the exact same multi-purpose payload, with all four modules running simultaneously:
Swapping Crypto Wallet Addresses: Because crypto addresses are long strings of random characters, users copy and paste them rather than typing them out. The script scanned pages, form fields, links, outgoing network requests, and even your device's clipboard, silently replacing legitimate wallet addresses with the attacker's own. Because the swapped addresses were valid wallets, they passed built-in checksum typo checks without raising alarms.
Skimming Payment Cards: Added in the afternoon, this module monitored form fields for credit card numbers, expiration dates, and three-digit CVV codes. However, it had a major blind spot: retailers that use embedded checkout boxes from payment processors like Stripe or Adyen load those fields inside isolated iframes, which scripts on the main page cannot read. Only cards typed directly into a website's own native form fields were exposed.
Stealing Cookies and Screenshots: Whenever the script swapped a crypto address near keywords like "wallet" or "payment," it snapped a screenshot of your browser window and exfiltrated it along with the page URL and accessible browser cookies. (Properly secured session cookies flagged as
HttpOnlyremained locked and out of reach.)Deploying iOS Exploit Kits: If you visited an affected site on an iPhone or iPad, the script checked your iOS version and loaded a hidden frame attempting to break into the device itself using two notorious exploit kits: Coruna and DarkSword.
Coruna targets iOS 13 through 17.2.1 using 23 chained exploits. Google's Threat Intelligence Group previously tracked it moving from commercial surveillance customers to state-backed spies and eventually cybercriminals.
DarkSword targets iOS 18.4 through 18.7. Published in March 2026 by Google, Lookout, and iVerify, a newer version was leaked publicly online days later—putting nation-state-grade mobile spyware tools onto mainstream news and banking websites.
Apple patched every flaw used by both kits in iOS 18.7.3 (released December 2025) and iOS 26.3(released February 2026). On newer iPhones, the script's own logs showed
weaponized = false, meaning it only gathered device details. While Hudson Rock notes there is no public proof of a real-world iPhone compromise from this wave, visitors on iOS 18.6 or older may have noticed an eerie clue: the webpage automatically refreshing itself every 60 seconds to retry the exploit.

Follow the Money—and the Trail
Because cryptocurrency ledgers are public, researchers could watch the payouts in real time:
By Monday evening, Report URI tracked roughly $7,040 sent to the attacker's wallets across Bitcoin, Litecoin, Tron, Ethereum, and BNB Chain. Step Security tallied roughly $6,600, while Cointelegraphcounted $7,100.
Alex Thorn of Galaxy Digital noted that only about $2,300 appeared to come from actual victims, with the rest looking like the attacker cycling their own funds.
Between 4:00 AM and 5:00 AM on Tuesday, the wallets were emptied. One cash-out address matched infrastructure Report URI had previously tied to attacks in March 2026 and November 3, 2026.
How many credit cards were skimmed or cookies stolen remains invisible on any public ledger.
How Did They Get In? (And It Wasn't Just Optimizely)
Early Tuesday, Optimizely confirmed on its status page and in a statement to BleepingComputer that an unauthorized party exploited a vulnerability in the Experimentation account signup flow—the portal where new users register accounts. Optimizely stated it blocked the vulnerable path, terminated active sessions, revoked access tokens, reset affected accounts, and notified impacted customers.
How a bug in new account signups allowed an attacker to edit existing enterprise customer projects remains an unanswered question awaiting Optimizely's full root-cause report.
Crucially, Optimizely wasn't the first door the attacker opened. Three days earlier, on November 13, Hudson Rock identified identical malicious loaders injected into three smaller web services: analytics tool Kissmetrics, story widget StorifyMe, and a video script from Minute Media (minute-ly). Through those tools, the malware reached a bank in the Philippines, the German newspaper Frankfurter Allgemeine, The Sydney Morning Herald, online retailer vidaXL, and Fox Sports.
A researcher running the security site Ohey discovered the exact same tracking channel ID across all four waves, suggesting a single buyer of the exploit kit ran the entire campaign—though Hudson Rock cautions that shared infrastructure does not definitively prove a single operator. While Ohey noted the iOS exploit kit's control panel was written in Chinese and used servers in Hong Kong and Singapore, researchers warn that digital breadcrumbs like language and server locations are trivial to spoof.
Why Website Defenses Failed
Many major websites rely on a Content Security Policy (CSP)—a rulebook telling your browser which domains are allowed to run scripts and where data can be sent. Report URI audited the CSP headers of the affected sites and found glaring gaps:
68 sites had no CSP at all.
50 sites had a CSP that was too loose, allowing the malware to both execute and exfiltrate stolen data.
17 sites allowed the script to run (meaning crypto address swapping still worked) while blocking data theft.
Only 24 sites (15%) had policies strict enough to block the attack completely.
The core problem is that a CSP checks where a file comes from, not what is inside it. Because every affected website had explicitly whitelisted cdn.optimizely.com, browsers welcomed the poisoned file right through the front door. Worse, Optimizely’s documentation notes that certain features require websites to enable unsafe-eval—a relaxed security setting that allows browsers to execute code constructed from plain text strings, which enabled the hidden backup copy to run on HSBC's site even when the external script was blocked.
While Subresource Integrity (SRI) checks a file's cryptographic fingerprint to ensure it hasn't changed, the Payment Card Industry (PCI) acknowledges SRI is impractical for dynamic A/B testing scripts that change every time a marketer tweaks a button. And while updated PCI rules require strict script monitoring on payment pages, an A/B testing snippet running on a homepage or news article falls completely outside those rules.
This mirrors past supply-chain disasters: the 2018 Ticketmaster UK breach via an Inbenta chatbot (which exposed 9.4 million European customers and drew a £1.25 million fine) and the 2024 Polyfill.io takeover (where security firm Sansec found malicious code injected into over 100,000 websites after the domain changed hands, leading to U.S. Treasury sanctions against Funnull).
What You Should Do Now
For Everyday Internet Users:
Check Crypto Transfers: If you copied or sent cryptocurrency while browsing any major news, tech, or retail site on Monday, November 16, verify your recent transaction history. Always verify every single character of a wallet address before sending—and if you use a hardware wallet like Ledger, trust onlythe address displayed on the physical device's screen.
Monitor Bank Statements: If you typed a credit or debit card number directly into a website's checkout form on Monday, watch your statements closely and contact your bank immediately if you spot unfamiliar charges.
Update Your iPhone or iPad: Ensure your device is running iOS 18.7.3 or iOS 26.3 (or newer), which patch every vulnerability targeted by Coruna and DarkSword.
For Website Owners and Security Teams:
Audit Third-Party Scripts: Monitor the actual contents and behavior of third-party JavaScript, not just the domains hosting them.
Tighten Your CSP: Restrict
connect-srcandscript-srcdirectives tightly, removeunsafe-evalwherever possible, and enable CSP violation reporting so visitors' browsers alert you when unexpected scripts attempt to load.Rotate Credentials: Audit and rotate API keys and user permissions for marketing and experimentation platforms.
Most of us think of a website as a single entity: one company, one address, one padlock. In reality, according to the HTTP Archive, the median major website stitches together 66 outside services directly inside your browser. On Monday, one text box inside just one of those trusted tools reached 178 organizations in a single day.
(News Worthy)
Next Article

Zero Trust: Inside the FBI’s Urgent Warning on ARTEX AI Hacking Agents
An FBI FLASH report warns that attackers are using the open-source ARTEX AI agent framework to breach networks in hours. Variedy breaks down how to defend.
10 Min

The AI Clone Market: Transforming Business Communication and Media Production
2 min read
