On November 16, 2026, visitors to the official websites of Zoom, the BBC, CNN, Fox News, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM and another 150 or so major brands had malicious code running in their browsers. In all, 178 organizations were caught up in the attack.
The script swapped crypto wallet addresses for the attacker’s on the fly, skimmed card numbers and CVVs straight from payment forms, sent cookies and screenshots of pages back to the attacker, and loaded exploit kits on iPhones and iPads.
Once again, Ledger, the crypto market’s leading purveyor of “hardware security,” made the honor roll.
The attacker never broke into any of those 178 companies directly. The break-in was at a single outside supplier, Optimizely, whose service lets companies run A/B tests on their websites. Write this off as “vendor error,” or blame “the marketing people who wanted to move a few buttons around on the storefront,” and once again the conversation steers away from the core disease of the entire tech industry.
The marketing people had nothing to do with it. A marketer doesn’t design security for a bank or a crypto wallet, configure browser security headers or decide whose code gets to run unchecked on a page where a customer types a card number or copies a crypto address.
The CEO, the CTO and the lead engineers make those calls. November 16 showed us just how well they understood what they were selling. The people running companies that sell security and financial infrastructure had only a surface grasp of their own products from day one.
When the Cook Makes the Rifles
Plenty of companies take on fiendishly difficult technical problems under bosses who never got past the sales brochure.
Picture a defense minister who never spent a day in the military, or a cook who makes sniper rifles and can’t shoot one.
You can drive a Toyota for five years and hold forth on cars without ever having rebuilt an engine or knowing what goes on under the hood. We all know how to switch on the kettle in the morning. Knowing how to boil water doesn’t make you an electrical engineer.
That’s how much of the crypto industry and corporate IT got started. Look at the rookie security holes that keep turning up in the same crypto wallets year after year. It’s hard to shake the feeling that their founders first heard about the next big thing over a beer one evening. They heard there were hundreds of millions to be made from people’s fear of losing their crypto, put up the money, hired equally shallow people and started selling the world “bulletproof security.”
The problem isn’t that somebody missed a single line of code. The vulnerability was baked into the choice of technology and the architecture from the start. A CEO and a lead engineer who understood how the web works and what threats they faced would never have let this setup anywhere near a customer. An outside server could send arbitrary JavaScript to a customer’s browser at any time, without anyone at the company knowing. That code would have full access to the page.
How the Attacker Walked In Through Optimizely
The technical side of this story is almost laughably simple. Researchers at Report URI laid it out in detail, and Optimizely itself confirmed it.
Optimizely gives its customers a script to include on every page so they can change their site’s interface without shipping a new release. Its dashboard has a built-in field called Project JavaScript. Whatever code a customer enters there, the official cdn.optimizely.com automatically serves to every visitor’s browser, where it runs on the customer’s site.
On November 16, the attacker found a flaw in Optimizely’s own account registration system. In its official notice, the company acknowledged that the flaw let someone act on behalf of the administrators of customer projects without even knowing their passwords.
Once inside 178 companies’ projects, the attacker added a tiny loader to Project JavaScript. It fetched the main malicious script, track.js, from an outside domain, 1688360.cc. One version concealed the address in base64. Another carried a backup copy of the malware inside the snippet itself. If the external domain was blocked, that copy ran through eval.
To a visitor’s browser, everything looked impeccable. The script arrived from the trusted cdn.optimizely.com over HTTPS with a valid certificate. Once on the page, it unpacked a whole arsenal:
- Crypto address swapping across the board. The script, known as a clipper, found addresses for Bitcoin, Ethereum and every EVM network, TRON, Solana, TON, Litecoin, Bitcoin Cash and Dogecoin, and replaced them with the attacker’s own. It even checked the checksums so the replacement address would be valid and the wallet wouldn’t flag an error. It swapped addresses in the text of the page, in input fields, in the clipboard when someone clicked “Copy,” and even in outgoing
fetchandXMLHttpRequestrequests. The malware intercepted JavaScript’s own built-in calls to get around the page’s defenses. - Skimming payment cards. When a customer entered a card number, expiration date and CVV into a payment form on the site itself, outside the isolated frame of a payment provider such as Stripe or Adyen, the script read the details and sent them to the attacker’s server.
- Session theft and surveillance. Every time the malware successfully swapped a crypto address, it exfiltrated the browser cookies and the page’s full URL, and took a screenshot of the victim’s screen.
- iOS exploits in the wild. When someone visited an infected page on an iPhone or iPad running iOS 18.6 or earlier, the script loaded the Coruna and DarkSword exploit kits to break into the mobile operating system.
The malicious code stayed on the websites of some of the world’s biggest brands from morning to evening on November 16. Report URI estimates that the attacker collected about $7,040 in the first nine hours from direct transfers to the substituted crypto addresses alone, and had moved the funds out by the morning of November 17. The haul was modest only because most of the victims on the list were news outlets and retailers. Every one of those blockchain transfers was irreversible.
Architectural Blindness Left the Door Open at 85% of the Giants
The most telling numbers in Report URI’s account came from checking the affected companies’ Content Security Policy (CSP) headers. CSP is a basic safeguard on the modern web. A site tells the browser which domains it can load scripts from and where those scripts can send data.
Even if a supplier like Optimizely gets hacked, a properly configured CSP will stop its script from fetching track.js from an unrelated domain like 1688360.cc or sending stolen card details or cookies off-site.
Report URI checked 160 domains belonging to the affected organizations.
- 68 companies had no CSP header at all.
- 50 companies had configured their policy so incompetently that it let the malware in and the stolen data out.
- Only 24 organizations (15%) blocked the attack outright.
The bank HSBC produced a masterpiece of its own. Its CSP blocked the external file from 1688360.cc. The attacker had a backup. The copy of the malware embedded in the Optimizely snippet ran on the bank’s website through eval().
Why would one of the world’s biggest banks allow unsafe-eval, an extremely dangerous flag that lets any text string run as code? Because, as Report URI points out, Optimizely’s own official documentation recommends that customers enable unsafe-eval!
Think that through. Engineers at a bank and some 150 corporations put an outside script on their live pages. Someone else can change what it does on the fly, from a cloud dashboard, without anyone reviewing the change. Then the engineers themselves switch off the browser’s protection against eval. This is what happens when the person who knows how to switch on the kettle gets to design the web infrastructure.
Ledger Claims the Moral High Ground With a Titanium Safe and a Wide-Open Front Door
Among the broadcasters and the furniture sellers, Ledger really stands out (Cointelegraph). Here’s a company whose entire reason for being boils down to “Your computer is dangerous. Keep your crypto with us.” On November 16, its official website, ledger.com, was serving a script designed to swap crypto addresses, steal card details and break into iPhones.
Ledger has spent years turning everything around its wallet into a sieve.
- In 2020, Ledger left the home addresses, phone numbers and names of hundreds of thousands of its wallet buyers exposed through a hole in its marketing and e-commerce systems.
- In December 2023, attackers hijacked the Ledger Connect Kit library that Ledger distributed through the third-party CDN
jsDelivr. Malicious JavaScript landed right inside decentralized apps and drained hundreds of thousands of dollars from users’ wallets. - In October 2026, we reported in detail on buyers who received modified Ledger devices with a hardware implant hidden inside the casing.
- And on November 16, 2026, there was
ledger.comagain, feeding its visitors a Trojan delivered by Optimizely’s A/B testing service.
You could predict the response from Ledger Donjon, the company’s security team, down to the last comma. Ledger confirmed the infection on ledger.com, then briskly announced that “the incident did not affect the architecture of the hardware devices, Ledger Live or users’ private keys.”
This is textbook hypocrisy. Customers can hardly reach Ledger’s “secure chip” by telepathy. Before they ever get their hands on the device, they visit the official ledger.com site to place an order and type their card details and home address into a form, download the app or look something up in the documentation.
And right on the wallet maker’s doorstep, there’s track.js to greet them. It skims the card details they enter, swaps addresses in their clipboard and tries an exploit against their iPhone.
Imagine telling a customer how secure the safe inside your device is while visitors are getting robbed at the front door of your own website, all because you wired in a third-party script you didn’t know how to handle. That isn’t cybersecurity. It’s a signed confession of total professional incompetence from whoever made the architectural decisions.
The Death of the “Official Website” and the False Comfort of an App Store Download
For decades, the first rule of staying safe online went like this: “Check the address bar, look for the HTTPS padlock and only visit the brand’s official website.” On November 16, that advice finally broke down. Visitors went to the genuine hsbc.com, zoom.us or ledger.com, complete with a spotless certificate, and wound up with somebody else’s wallet address in their clipboard, courtesy of code the site owners had never even seen.
The modern corporate website is no longer the company’s fortress. It’s a shared apartment, and its inept architects have let twenty outside contractors move in. A/B testing services, advertising pixels, chat widgets, external fonts and trackers all live there. Compromise any one of those tenants and a thief sets up shop under the bank’s or the wallet maker’s own sign.
Nor are you any safer these days with the naive line “I don’t use websites. I’ve got the official app from the App Store or Google Play.” A huge share of the crypto wallets and fintech apps in those stores are those very same web pages wrapped in a WebView.
Even the hybrid or native apps come stuffed with third-party analytics SDKs and mechanisms that quietly fetch fresh JavaScript from a server without going through app store review. If the app’s architect follows the same “let’s plug in a dozen third-party cloud scripts and let them do whatever they like” philosophy, the pretty icon on your phone becomes the same trap as a browser tab.
The Optimizely attack isn’t a story about genius hackers. The story is an industry that traded security for marketing slides long ago. The people at the helm of companies entrusted with other people’s money know how to switch on the kettle. They haven’t the faintest idea where the electricity in the socket comes from.