Protego field desk
Cybersecurity11 min read

The Optimizely Hack Explained: How 178 Brands Served Malware

On November 16, 2026, an attacker abused a flaw in Optimizely’s signup flow to inject malware into the A/B testing scripts of 178 organizations, including the BBC, CNN, Zoom and Salesforce. Here is what Optimizely is, how the attack worked, what was stolen and what to do now.

I
Microsoft Cloud Solution Architect

Quick answer

On November 16, 2026, an attacker abused a flaw in Optimizely’s account signup flow to add malicious JavaScript to customer A/B testing snippets. For about 19 hours, sites of 178 organizations served code that swapped crypto addresses, skimmed cards, stole cookies and targeted older iPhones. About $7,000 in crypto was traced.

A trusted website receiving injected projectJS code from a CDN, leaking card numbers, swapped crypto addresses, cookies and iPhone exploits
A trusted website receiving injected projectJS code from a CDN, leaking card numbers, swapped crypto addresses, cookies and iPhone exploits
Optimizelysupply chain attackthird-party script securityMagecartcard skimmingContent Security PolicyiOS exploit

Picture this. On Monday, November 16, 2026, you open the BBC website on your iPhone, or log in to your bank, or copy a crypto deposit address from an exchange’s help page. Nothing looks wrong. The site is the real site, the padlock is there, the address bar is correct. But for most of that day, a script running quietly inside hundreds of major websites was watching what you typed, swapping crypto addresses you copied, and on older iPhones, trying to break into the phone itself.

None of those websites were hacked in the usual sense. Their servers were fine. The malicious code came from a tool they all trusted: Optimizely, a marketing platform used to run A/B tests. This article explains what Optimizely is, how the attacker got in, what was stolen, how much, who was affected, and where things stand now.

The Optimizely attack in numbers

Figures from Report URI’s investigation and BleepingComputer’s wallet analysis, as of November 17, 2026.

178
Organizations whose Optimizely snippet carried malware
Report URI
188 of 1,611
Optimizely snippets found infected out of those checked
Report URI
~19 hours
From first sighting (04:21 UTC) to fully clean (23:11 UTC)
Report URI
~$7,000
Crypto received by the attacker’s wallets
BleepingComputer, Report URI

What is Optimizely, and why is it on so many websites?

Optimizely is a digital experience and experimentation platform. Its Web Experimentation product lets marketing teams change a live website without asking developers: test a green "Buy" button against a blue one, try a new headline, show a different offer to half the visitors, then measure which version wins. That is called A/B testing.

To make that work, a company adds one line of code to every page of its site. That line loads a small JavaScript file, called the snippet, from Optimizely’s servers at cdn.optimizely.com. Every time anyone visits the page, their browser downloads the latest snippet and runs it, with the same powers as the website’s own code. It can read the page, change it, and see what you type.

That is the whole point of the product, and also the whole risk. Whoever can change the snippet can change every page of every site that loads it, instantly, without touching the company’s own servers. The snippet even has a field built for exactly this: Project JavaScript (projectJS), an optional box where a customer can paste custom code that runs on every page load.

What happened, in one paragraph

On November 16, an attacker gained the ability to edit customer projects inside Optimizely and pasted a malicious loader into the Project JavaScript field of hundreds of them. Optimizely’s content delivery network then did exactly what it is designed to do: it served the updated snippet, with a valid certificate, to every visitor of those websites. Security researcher Scott Helme of Report URI, whose company monitors scripts on websites, found 188 infected snippets belonging to 178 organizations after checking 1,611 of them, and says the true number is likely higher. As Report URI put it, Optimizely’s network "served exactly what it was given". This was not a hack of the CDN itself.

How did the attacker get in?

This is the part with the least certainty, so it is worth being precise about who said what.

  • Optimizely’s explanation: in a statement to BleepingComputer, Optimizely said "an unauthorized party exploited a vulnerability in the Experimentation account signup flow". An email to customers, quoted by CSO Online, adds that the attacker used it "to impersonate a user and add malicious JavaScript to some customer projects".
  • What that does not explain: a flaw in the signup page should let you create your own account, not edit other companies’ projects. Optimizely has not said how one led to the other. CSO Online also reports Helme’s view that a feature for adding members to a customer’s team was used to plant the code, a detail the two accounts do not reconcile.
  • A timeline gap: according to analyst Frank Dickson, quoted by CSO Online, Optimizely places the first malicious change at 07:51 UTC. Report URI has a capture of the BBC’s snippet loading attacker code at 04:21 UTC, three and a half hours earlier. That gap is unexplained so far.
  • Still active during cleanup: Smartsheet’s snippet was cleaned and then infected again with a disguised version, which Report URI reads as the attacker still having access while companies were cleaning up.

The short version: a weakness in how Optimizely let people into its platform gave an outsider write access to real customer projects. The exact mechanics are not public yet.

How malicious code reached millions of visitors

No affected company’s own servers were touched. The trusted third-party script did the delivery.

  1. Step 1: Abuse the signup flaw

    The attacker exploits a vulnerability in Optimizely’s account signup flow and impersonates a user with access to customer projects.

  2. Step 2: Edit Project JavaScript

    A small loader is pasted into the projectJS field of hundreds of customer projects. The rest of each snippet stays untouched.

  3. Step 3: CDN serves the update

    cdn.optimizely.com delivers the modified snippet with a valid certificate, exactly as designed.

  4. Step 4: Visitor’s browser runs it

    Anyone opening an affected page runs the loader, which pulls the real payload from 1688360.cc/track.js.

  5. Step 5: Steal and exploit

    The payload swaps crypto addresses, skims cards, grabs cookies and screenshots, and targets older iPhones.

The loader came in three flavors, which tells you the attacker expected defenses. One was a plain script tag pointing at https://1688360.cc/track.js. One hid that same address in base64 so a simple text search would miss it. The third carried a full copy of the malware inline and ran it with eval, so it would still work on sites that blocked the attacker’s domain. Report URI counted 26 rebuilds of the payload during the day as new features were added.

What did it steal, and how?

The payload was four tools in one. Here is what each did, in plain terms.

AttackWhat happened to the visitorWhere the data went
Crypto address swappingAny wallet address on the page, in a form, or copied to the clipboard (Bitcoin, Ethereum, TRON, Solana, TON, Litecoin, Bitcoin Cash, Dogecoin) was silently replaced with the attacker’s own. Send money to "your" address and it went to them.Straight to attacker wallets
Card skimming (from 13:43 UTC)When a form or background request contained a valid card number, expiry date and CVV, the script copied them.1688360.cc/card.php
Cookie and screenshot theftWhen it swapped an address near wallet or payment wording, it sent the page address, the readable cookies (which can include login sessions) and a screenshot of the page.1688360.cc/hit.php
iPhone and iPad exploitsOn iOS devices it loaded exploit kits in a hidden frame, chosen by iOS version: one called Coruna for iOS 13 to 17, one called DarkSword for iOS 18.x, and a fingerprinting chain for iOS 26.Attacker servers on throwaway .icu domains

Two details matter for anyone worried about their card. First, the skimmer could only read card details typed into the website’s own forms. Payment boxes hosted by Stripe, Adyen and similar providers run in a separate, isolated frame that the script could not reach. Second, the skimmer was only added at 13:43 UTC, so card details entered earlier in the day were not exposed this way.

The iPhone part is the most serious, because it does not need you to do anything. Report URI says code comments in the kit treat iOS 18.7.3 and later, and iOS 26.3 and later, as patched. On iOS 18.6 or earlier, affected pages reloaded themselves every 60 seconds to retry the exploit, which was the one visible symptom. Report URI also says the Coruna loader was byte-for-byte identical to a sample it had analyzed in October from a different campaign.

How much was stolen?

The crypto part is the only piece anyone can count, because blockchains are public. According to Report URI and BleepingComputer, the attacker’s wallets received roughly $7,000 to $7,150, depending on when you take the snapshot:

CurrencyAmount receivedApprox. USD
Bitcoin0.08528 BTC$5,400 to $5,500
Litecoin17.02 LTC$760
TRON (USDT + TRX)662.9 USDT + 183.7 TRX$719
Ethereum0.0553 ETH$100
BNB Chain (USDT)82 USDT$82

Two caveats. BleepingComputer notes some of these wallets were active before the attack, so not every dollar is necessarily from this campaign. And the attacker moved the funds out early on November 17, between 04:11 and 05:39 UTC. Report URI links one of the cash-out addresses to earlier cash-outs in March and on November 3, 2026.

The real damage is probably much larger, it just cannot be measured from outside. Nobody outside the attacker knows how many card numbers, session cookies or screenshots were captured, or how many iPhones were successfully compromised. Report URI says plainly that it cannot see those numbers.

Who was affected?

BleepingComputer confirmed through urlscan.io captures that these sites loaded, or attempted to load, the malicious code: Zoom, KLM, GitLab, BBC, Scheels, CNN, American Girl, ManageBuilding, Smartsheet, Stamps.com, Salesforce, Fox News, Mutual of Omaha, American Eagle, Air France, Change.org, Optimizely itself, HSBC Insurance, Bitly, IG and UNICEF USA.

Report URI and CSO Online name further organizations, including IKEA, HSBC, Equifax, Paychex, Starbucks, Levi’s, Costco, Heineken, Mattel, Peacock, Vodafone Australia, Virgin Media and Ledger, a company that makes hardware crypto wallets. Even Optimizely’s own www.optimizely.com and login.optimizely.com snippets carried the code. Some companies host the Optimizely snippet on their own domain; those were affected too, because the content came from the same project.

By sector, Report URI’s breakdown of the 178 organizations looks like this:

SectorOrganizations
Banking and finance32
Technology and SaaS23
Retail and fashion20
Streaming12
News and media10
Other sectors81

Being on this list does not mean a company’s systems or customer databases were breached. It means visitors to its website were served malicious code for part of November 16. That distinction matters, but for a visitor who lost money it is not much comfort.

Timeline of the attack (UTC)

Based on Report URI’s timeline, Optimizely’s statement and BleepingComputer. Dates are 2026.

  1. Earlier test wave

    The same actor injects a loader through other third-party scripts (Kissmetrics, StorifyMe, Minute Media). Not counted in the 178.

  2. First Optimizely injection seen

    BBC’s Optimizely snippet is captured loading attacker code. Optimizely’s own timeline reportedly starts at 07:51.

  3. Main payload live

    1688360.cc/track.js starts loading on bbc.com. The domain was registered hours earlier.

  4. First crypto sweep

    The attacker moves stolen funds to exchanges.

  5. Card skimmer added

    The payload starts capturing card numbers, expiry dates and CVVs.

  6. iOS exploit chain expanded

    A second iOS server is added; every visit from iOS 18.6 or earlier gets an exploit attempt.

  7. Report URI finds the infection

    A sweep of Optimizely snippets turns up the infected files.

  8. Mass cleanup

    Ledger, Smartsheet, BBC and others are cleaned. One file (TalkTalk) is still infected at 21:52.

  9. All files clean

    Report URI confirms no infected Optimizely snippets remain.

  10. Optimizely status notice

    Optimizely posts a security incident notice and says the vulnerability is fixed.

  11. Wallets emptied

    The attacker cashes out the Litecoin, USDT and Bitcoin.

Where does it stand now?

As of November 18, 2026:

  • The malicious code is gone. Report URI confirmed all infected Optimizely files were clean by 23:11 UTC on November 16.
  • Optimizely says the hole is closed. Its status page says it fixed the vulnerability, removed the injected code and secured affected accounts. Per Search Engine Journal and CSO Online, it also ended active sessions, revoked tokens, reset access for affected users, blocked unauthorized code from being delivered at the Cloudflare level, and notified all affected customers. The incident was still marked "Monitoring".
  • Customers are reacting. Salesforce told BleepingComputer it is investigating and has disabled Optimizely scripts across its preview and production instances, and that it will notify customers if it finds unauthorized access to their data.
  • No one has claimed it. Helme told CSO Online the attackers are "either Chinese or they really want us to think that they are Chinese". That is an informed read of the evidence, not a formal attribution.
  • Open questions: how a signup flaw gave write access to other customers’ projects, why the timelines differ by three and a half hours, how many cards and sessions were stolen, and how many iPhones were compromised. No regulator or lawsuit had been reported yet.

How to tell if you were affected

You were only exposed if you opened one of the affected sites on November 16, 2026, roughly between 04:00 and 23:00 UTC. If that might be you, check each risk separately, because each one leaves a different trace:

RiskHow to checkSign that you were hit
iPhone or iPad exploitSettings > General > About > iOS VersionYou were on iOS 18 below 18.7.3, or iOS 17 or older. An affected page that kept reloading itself about once a minute is a strong sign the exploit ran
Crypto address swapOpen the transfer in your wallet or exchange history and compare the destination with the attacker wallets listed belowThe destination matches one of the attacker addresses, or differs from the address the recipient gave you
Card skimmingThink back: did you type a card number into a form on the site itself after 13:43 UTC?Yes, and it was not a Stripe, Adyen or similar hosted payment box. Watch for charges you do not recognize
Session cookie theftCheck the active sessions or devices list in the account settings of the affected siteA session from a device or place you do not recognize

These are the attacker wallets Report URI published. If you sent crypto on November 16, search your transaction history for them:

BTC   19g9rZcPeiz8rUvyWpTRyXXhPnjsWrc3kW
ETH   0xC4026b06D82b0F0B363eda0642D61CfbDc17884e   (also BNB Chain and other EVM chains)
TRON  THvs6p56XkUiMkJeSc51cAP64u2RsTYqxa
SOL   FVtL8NMYS9PA1v3HE67XUGCtGA9JaZUvqav2fU2LrV5n
TON   UQAgM1VqSZQiJLSG5w-ovurq5CZdxLEK3_7CMmwALTA5e09p
LTC   ltc1q3wp7g6ds0ggg6fl6kncpa4sepn0q3qfacgred5
BCH   qzkaaf8n6grmpr0d9lx8xutn3r9lk722f566hhpe58
DOGE  DBEGrxRF7KMSuhTS4c1MFviStU6efdMnu5

A crypto transfer to one of these addresses cannot be reversed by your wallet, but report it to the exchange you used and to your local police cyber unit: exchanges sometimes freeze funds that reach them.

What to do as a visitor

If you visited any of the sites above on November 16, 2026, especially between about 04:00 and 23:00 UTC, here is what is worth doing:

  1. Update your iPhone or iPad. If you are on iOS 18 and below 18.7.3, or iOS 26 and below 26.3, update now. This is the most important step, because the exploit did not need any action from you.
  2. Check any crypto transfer you made that day. Compare the address you sent to with the address the recipient actually gave you. If you pasted an address that day, the clipboard swap may have changed it.
  3. Watch your card statements. If you typed card details directly into a form on one of these sites after 13:43 UTC on November 16, keep an eye out for charges you do not recognize, and ask your bank about a replacement card if you are unsure.
  4. Log out and back in on sensitive accounts. Stolen cookies can work like a temporary key to your session. Signing out everywhere ends those sessions. If you are not sure whether your details have ended up somewhere they should not be, our free password breach check is a quick place to start.

If you run a website: what this teaches about third-party scripts

The uncomfortable lesson is that the affected companies did nothing unusual. They added a popular, trusted marketing tool the way its documentation told them to. The attack still ran on their pages, under their brand, against their customers. Any script you load from someone else, whether A/B testing, analytics, chat widgets, tag managers or ad code, runs with the same powers as your own code. If they get hacked, you get hacked.

This has happened before. In 2018, attackers modified a third-party library on British Airways’ payment page and skimmed customer card details, the best-known "Magecart" case. In 2024, the polyfill.io domain changed hands and began serving malicious code to more than 100,000 websites that loaded it. Our guide to detecting and removing card skimmers covers the same family of attacks from the store owner’s side.

What actually helped during this incident was a Content Security Policy (CSP): a header that tells browsers which domains your page may load scripts from and send data to. Report URI and Search Engine Journal give the breakdown for 159 of the affected organizations’ own sites:

CSP status on the affected siteOrganizationsWhat happened
No CSP at all68Payload ran and data could be sent out freely
CSP that allowed it anyway50Payload ran and data could be sent out
CSP limited where data could go (connect-src)17Payload ran, but stolen data could not leave
CSP blocked the payload24 (15%)Attack stopped in the browser

HSBC is the instructive case: its CSP blocked track.js and blocked the data theft, but the inline fallback copy still ran because the policy allowed unsafe-eval. CSO Online and Search Engine Journal both point out that Optimizely’s own CSP guidance recommends allowing 'unsafe-eval' and 'strict-dynamic', which is what weakened several policies.

Were you one of the 178? How to check

If your site uses Optimizely Web Experimentation, do this first:

  1. Check your inbox. Optimizely says it notified every affected customer directly.
  2. Review your project history. In Optimizely, look at the change history for Project JavaScript and for every experiment’s custom code on November 16. Any edit you cannot tie to a person on your team is the attacker.
  3. Search your logs and CSP reports. Look for requests to 1688360.cc or any of the domains in the indicators section below. A CSP violation report for one of them means the payload tried to run on your pages.
  4. Confirm the snippet is clean now. Find your snippet ID and check what it serves today. This proves the cleanup, not whether you were hit on the day:
# 1. Find the Optimizely snippet your homepage loads
curl -s https://www.example.com | grep -oE 'cdn\.optimizely\.com/js/[0-9]+\.js'

# 2. Check that snippet for the known loader patterns (no output = clean)
curl -s https://cdn.optimizely.com/js/YOUR_ID.js | grep -E '1688360|webshell\.lol|atob\('

If you were affected, treat it as a breach of your site even though your servers were untouched: card numbers typed into your own forms after 13:43 UTC and any session cookies readable by JavaScript should be considered exposed, and depending on where you operate you may have a duty to notify customers.

Prevent the next one

A practical checklist:

  1. Inventory every third-party script on your site. Include the ones added through a tag manager. You cannot protect what you do not know you load.
  2. If you use Optimizely, audit now. Check Project JavaScript and every experiment’s custom code for 1688360.cc, atob( or anything you did not write. Review users, API tokens and integrations, and rotate access.
  3. Lock down where data can go. A strict connect-src is the cheapest control here: in 17 cases it let the malware run but stopped it from sending anything home. Start in report-only mode so you can see what would break.
  4. Avoid unsafe-eval wherever you can. It is what let the inline fallback run on sites that had otherwise blocked the attacker’s domain.
  5. Keep payment fields in your provider’s hosted frame. The skimmer could not read Stripe or Adyen hosted fields. Cards typed into your own form fields were exposed.
  6. Treat marketing tool access like production access. Whoever can edit your A/B testing or tag manager project can effectively deploy code to your site. Require MFA, limit who has publish rights, and review the access list.

Here is what a tighter policy looks like in principle. The exact domains depend on your site, so treat this as a starting shape, not a copy-paste fix:

Content-Security-Policy-Report-Only:
  default-src 'self';
  script-src 'self' https://cdn.optimizely.com;
  connect-src 'self' https://logx.optimizely.com;
  frame-src https://js.stripe.com;
  report-uri https://your-report-endpoint.example/csp

You can check what security headers your own site sends today, including whether it has a CSP at all, with our HTTP headers checker or the broader free website vulnerability scanner, which also lists the third-party scripts your page loads.

Indicators of compromise

For security teams searching logs or proxy data, these are the main indicators published by Report URI. Block them including subdomains:

1688360.cc              main payload and exfiltration (136.113.106.136)
5vobauua3ojpklx.icu     iOS exploit router (34.96.213.201)
eil786npto9u5bq.icu     iOS exploit router (34.87.36.58)
l7bcsnwc7nzr3uo.icu     exploit kit telemetry
webshell.lol            injector used in the earlier wave
je4ft3gn8vxudum.icu     fallback C2
q1ls1ho9x7wsxlu.icu     fallback C2
ios-hack.icu            fallback C2

track.js SHA-256 (latest build):
f7b84b9a94f749f8ced3f0112d0a66894e8bbc3b4aa5592021d1b8976a9bad40

Frequently asked questions

Was Optimizely itself hacked?

Partly. Optimizely says an attacker abused a vulnerability in its Experimentation account signup flow to impersonate a user and edit customer projects. Its content delivery network was not compromised; it delivered the modified files exactly as instructed. The affected websites’ own servers were not hacked either.

How do I know if I was affected by the Optimizely attack?

Only people who opened an affected site on November 16, 2026, roughly between 04:00 and 23:00 UTC, were exposed. Check four things: your iOS version (iOS 18 below 18.7.3 or older was targeted), any crypto transfer that day against the attacker wallets listed in this article, whether you typed a card number into the site’s own form after 13:43 UTC, and the site’s active sessions list for devices you do not recognize.

How many websites were affected?

Report URI found malicious code in 188 Optimizely snippets belonging to 178 organizations, out of 1,611 snippets it checked. Because it could only check snippets it knew about, it believes the real number is higher. Optimizely has not published its own count.

How much money was stolen?

Public blockchain records show roughly $7,000 reaching the attacker’s crypto wallets. The amount taken through stolen card numbers, session cookies and compromised iPhones is unknown, and is likely the bigger share of the damage.

I visited one of these sites on my iPhone. Am I compromised?

Not necessarily. The exploit targeted iOS versions before 18.7.3 and 26.3, and success is never guaranteed. The safest step is to update iOS now. If your phone was on an older version and you visited an affected site on November 16, also consider changing passwords for accounts you used on that device.

Is it safe to visit these websites now?

Yes, as far as is publicly known. All infected Optimizely files were cleaned by the night of November 16, Optimizely says the vulnerability is fixed, and some companies, like Salesforce, have switched Optimizely off entirely while they investigate.

Would Subresource Integrity (SRI) have stopped this?

Not in practice. SRI pins a script to an exact fingerprint, which works for files that never change. An A/B testing snippet changes every time a marketer edits an experiment, so it cannot be pinned. That is why a Content Security Policy, which controls what a script can load and where it can send data, was the control that actually made a difference here.

RD / 01

Reader desk

Discuss this guide

Rate the guide or ask a practical follow-up. Clear questions publish immediately; uncertain submissions wait for review.

Get the free Cloud Security Checklist

A 20-point hardening checklist for AWS, Azure, and GCP workloads. No spam, unsubscribe anytime.

I

Microsoft Cloud Solution Architect

Cloud Solution Architect with deep expertise in Microsoft Azure and a strong background in systems and IT infrastructure. Passionate about cloud technologies, security best practices, and helping organizations modernize their infrastructure.

Share this article

Licensed under CC BY 4.0: you may share and adapt this article with attribution to Protego.