# Optimizely Hack: The Cyber Attack That Hit CNN, BBC and Zoom

By [Eoin Villarin](https://amvprosox.com/about). Published 2026-11-17, updated 2026-11-18. Canonical: https://amvprosox.com/blog/optimizely-hack-cyber-attack-third-party-scripts

> Attackers abused Optimizely to push a card skimmer, crypto stealer and iPhone exploit kits to visitors of 178 major websites. Updated November 18 with new findings, company statements and how to protect your own site from third-party script attacks.

## What was the Optimizely cyber attack on CNN, BBC and Zoom?

On November 16, 2026, attackers abused a flaw in Optimizely Web Experimentation to add malicious JavaScript that ran on sites of at least 178 organizations, including CNN, the BBC, Zoom and HSBC. It swapped crypto addresses, skimmed cards from on-site forms, stole cookies and delivered iPhone exploit kits. Optimizely says the flaw is fixed. As of November 18, no attacker or iPhone compromise has been confirmed.

**Key takeaways**

- On November 16, 2026, malicious JavaScript was served through Optimizely to visitors of at least 178 organizations.
- The code swapped crypto addresses, skimmed cards from on-site forms, stole cookies and pushed iOS exploit kits.
- Optimizely says it fixed the flaw; as of November 18 no attacker or iPhone compromise has been confirmed.
- Only 24 affected organizations had a Content Security Policy strict enough to block the payload.
- Script inventories, a strict CSP, change monitoring and hosted payment fields are the main defenses.

If you searched for a cyber attack today, the big story is the Optimizely hack: on November 16, 2026, attackers abused a flaw in Optimizely Web Experimentation to inject malicious JavaScript into customer projects, and that code ran for most of a day on the websites of at least 178 organizations, including CNN, the BBC, Zoom, Salesforce and HSBC. The script swapped crypto wallet addresses, skimmed card details typed into on-site forms, stole cookies and screenshots, and tried to load iPhone exploit kits. Below is what happened, hour by hour, and a practical plan any site owner can use to defend against the same kind of third-party script supply-chain attack.

## Update, November 18, 2026: what is new

A day after disclosure, some questions have answers and others are still open. These are the facts verified since the first reports.

- **No confirmed iPhone compromise.** Hudson Rock, which analyzed the mobile stage, says its evidence does not establish a zero-day or a successful exploit on a real device, and it gives no victim count. Later DarkSword stages recovered by a researcher describe uploading device data and collecting wallet seed phrases, but Hudson Rock has not reproduced them, and an iOS 26.3 chain it found was marked as not weaponized in its own log.

- **Not only Optimizely.** Hudson Rock traced the same loader infrastructure through four delivery services: Optimizely, Kissmetrics, StorifyMe and the minute.ly video player. The earliest poisoned response it found is from November 13. How the attacker got into the other three is still unknown, and Hudson Rock does not claim a single operator.

- **No attribution yet.** No group has been named. Report URI says a TRON address used to cash out this attack was also used to cash out earlier operations in March 2026 and on November 3, 2026, which links it to other attacks in its data.

- **Stolen crypto moved.** Between 04:11 and 05:39 UTC on November 17, the attacker moved the LTC, part of the USDT and the BTC to new addresses. Report URI's total stands at about $7,151 in on-chain receipts.

- **Infrastructure down.** The earlier injector domain `webshell[.]lol` was put on hold at 03:37 UTC on November 17, and the `.icu` exploit hosts no longer respond. Hudson Rock warns that blocking domains is not enough on its own, because some poisoned scripts carried the full payload inline.

- **More company statements.** HSBC told Reuters the unauthorized scripts were not active on its websites, which differs from Report URI's finding that the inline fallback ran on an HSBC page. Ledger said on X that its website was affected, the code was removed and its hardware wallets were not compromised, as reported by 36Crypto. Salesforce told BleepingComputer it disabled Optimizely scripts across preview and production and will notify affected customers "without undue delay".

- **Optimizely is still monitoring.** As of November 18 its status page has a single update, no post-incident report, no count of affected customers and no explanation of how a signup flaw gave write access to other customers' projects. Its statement to reporters adds that it ended active sessions, reset access for affected accounts, blocked the code at Cloudflare and notified all affected customers. SecurityWeek notes this is separate from a voice phishing breach Optimizely disclosed in February.

- **Fuller list of sites.** Newsweek published all 178 organizations by sector, adding names such as NPR, ASOS, Mattel, Starbucks Canada, Alaska Airlines, Virgin Media O2, Vodafone Australia, PokerStars, Invisalign and Tableau. Being on the list means the script was served, not that visitor data was taken.

## What happened in the Optimizely hack?

Optimizely is an A/B testing and personalization platform. Sites add one JavaScript snippet, usually from `cdn.optimizely.com/js/<project ID>.js`, and marketers change pages from a dashboard. That snippet runs on every page with full access to forms and cookies.

According to the technical write-up from Report URI, the attacker wrote a small loader into the **Project JavaScript** field of affected projects. The rest of the snippet was untouched and still served from Optimizely's CDN with a valid certificate. Report URI says this was not CDN cache poisoning: the CDN served exactly what the project contained. Sites serving the snippet from their own domain were hit too.

Optimizely's status page says "an unauthorized party exploited a vulnerability in Experimentation" to add the code. A spokesperson later told reporters, as quoted by BleepingComputer, that the flaw was in the Experimentation account signup flow. As of November 18, Optimizely has still not explained how a signup flaw gave write access to unrelated customer projects. Its own `www.optimizely.com` and `login.optimizely.com` snippets also carried the code.

![Rows of black server racks with network cabling in a data center room](https://amvprosox.com/api/img/1794920313579-bmxpws)
*The malicious loader was delivered from the vendor's own trusted CDN, so nothing looked out of place at the network level. Photo: Carl Lender from Sunrise, USA, CC BY 2.0, via Wikimedia Commons*

## Which websites were affected?

Scott Helme's team at Report URI fetched 1,611 Optimizely snippets and found malicious code in 188 of them, belonging to 178 organizations. They say the real number is probably higher, because they could only check the snippets they found in time. Reported names include:

- **News and media:** CNN, BBC, Fox News, Peacock

- **Tech and SaaS:** Zoom, Salesforce, GitLab, Smartsheet, Hootsuite, Bitly

- **Finance:** HSBC, Equifax, Mutual of Omaha, IG, and crypto wallet maker Ledger

- **Retail and travel:** IKEA, American Eagle, American Girl, Scheels, Air France and KLM

- **Telecoms and others:** TalkTalk, Change.org, UNICEF USA, Stamps.com

By StepSecurity's count, banking and finance led with 32 organizations. Newsweek's sector breakdown of the same list agrees, followed by technology and software with 23 and retail and fashion with 20.

## What did the malicious script do?

The loader pulled a second-stage file called `track.js` from the domain `1688360[.]cc`. There were three loader variants: a plain script tag, a base64-hidden URL decoded with `atob()`, and a full inline copy that ran through `eval` if the domain was blocked. Report URI counted 26 builds of `track.js` in one day.

### Crypto address swapper

The script carried hard-coded attacker addresses for ETH, BTC, LTC, TRX, TON, SOL, BCH and DOGE. It only replaced addresses that passed the coin's checksum and sat near words like "wallet", "deposit" or "send". It rewrote page text, form fields, the clipboard and outgoing requests, so an address could change between copy and paste.

### Card skimmer

Added at 13:43 UTC, the skimmer hooked form submits, `fetch` and XHR, looked for card numbers that pass the Luhn check with a valid expiry and CVV, and beaconed them to `/card.php`. It could not read card fields inside a payment provider's cross-origin iframe, such as hosted fields from Stripe or Adyen.

### Cookie and screenshot theft

After a successful address swap, it loaded `html2canvas`, took a JPEG of the page and posted it with up to 8,000 characters of `document.cookie`. HttpOnly cookies were out of reach.

### iOS exploit kits

On iPhones and iPads, the script loaded `boot.js` from `.icu` domains and opened a hidden iframe. Researchers identified the Coruna kit (aimed at iOS 13.0 to 17.2.1) and DarkSword (iOS 18.4 to 18.7). Pages on iOS 18.6 or earlier reloaded every 60 seconds to retry. So far the evidence shows the exploit pages were delivered, not that any iPhone was confirmed compromised.

![Four iPhone 15 Pro and Pro Max models lying side by side, two showing the Apple logo on screen](https://amvprosox.com/api/img/1794920312647-taepj9)
*The kit's own comments list iOS 18.7.3 and later, and iOS 26.3 and later, as patched. Photo: Ka Kit Pang, CC BY-SA 4.0, via Wikimedia Commons*

## Timeline of the attack (UTC)

| Time | Event |
| --- | --- |
| Nov 13 to 16 | Earlier wave via a separate injector (webshell[.]lol) in Kissmetrics, StorifyMe and Minute Media scripts, per Report URI |
| Nov 16, 04:21 | First Optimizely injection seen, in a BBC snippet |
| Nov 16, 13:43 | Card skimmer added |
| Nov 16, 17:13 to 17:57 | Coruna iOS iframe added for iOS 18.6 and earlier |
| Nov 16, 19:25 | Attacker wallets had received about $7,040, per Report URI |
| Nov 16, 20:40 | Mass cleanup begins |
| Nov 16, 23:11 | All infected Optimizely files clean, per Report URI |
| Nov 17, 00:37 | Payload domain placed on registry hold |
| Nov 17, 01:02 | Optimizely posts its incident notice; status still "Monitoring" on Nov 18 |
| Nov 17, 03:37 | Earlier injector domain webshell[.]lol placed on hold |
| Nov 17, 04:11 to 05:39 | Attacker moves LTC, USDT and BTC to new addresses, per Report URI |

The money figures differ: Report URI later raised its total to roughly $7,151, while StepSecurity estimated about $6,600. The on-chain loss is the smallest part of the story, since a blockchain shows nothing about stolen cards or cookies. Helme has said there is no way for researchers to tell how many cards were skimmed.

## How did the companies respond?

Optimizely says it fixed the vulnerability, removed the code, secured affected accounts, revoked tokens and notified affected customers. Reuters reported that Salesforce disabled the Optimizely script changes "as a precaution", Equifax stopped using Optimizely and said it found no sign its own systems were compromised, and IKEA was checking for impact. Apple, Zoom and CNN had not responded at the time of the Reuters report. HSBC, Ledger and Salesforce have since given fuller statements, covered in the November 18 update above.

## Why a strict CSP would have stopped most of it

Report URI's crawl of affected hosts is telling. Only 24 organizations had a Content Security Policy strict enough to block the payload. 68 had no CSP at all, 50 had a policy that still let the script run and send data out, and 17 blocked the data theft through `connect-src` but not the in-page address swapping. Two settings let the attack through: `'strict-dynamic'`, which trusts any script a trusted script inserts, and `'unsafe-eval'`, which let the inline fallback run on HSBC's site even though its CSP blocked `track.js`. Report URI notes that Optimizely's own documentation tells customers to allow `'unsafe-eval'`.

## How to protect your website from third-party script attacks

Here is the order I would work through this week on any site that loads analytics, testing, chat or payment scripts.

### 1. Inventory every third-party script

Open your key templates (home, login, checkout) in Chrome DevTools, filter the Network tab by JS and list every external host, including what each one loads. Record the owner, the business reason, who can edit it in the vendor dashboard, and whether it runs on payment or login pages. If nobody can justify a script, remove it.

### 2. Lock down vendor dashboards

The attack lived in a dashboard field, not in your code. Review users and API tokens in Optimizely, Google Tag Manager and similar tools. Remove old staff and agencies, enforce SSO or MFA, use publish approval where the tool supports it, and export change history regularly. Optimizely customers should also read every Project JavaScript field and experiment code block for `1688360`, `atob(` and anything unfamiliar, since the base64 variant defeats a simple text search.

### 3. Deploy a Content-Security-Policy with a host allowlist

Use an explicit `script-src` allowlist and a tight `connect-src`, avoid `'unsafe-eval'` and `'strict-dynamic'` on sensitive pages, and turn on reporting. Shown on several lines for reading; send it as one header line:

```
Content-Security-Policy:
  default-src 'self';
  script-src 'self' 'nonce-{RANDOM_PER_RESPONSE}' https://cdn.vendor.example;
  connect-src 'self' https://events.vendor.example;
  img-src 'self' data: https://cdn.vendor.example;
  style-src 'self' 'nonce-{RANDOM_PER_RESPONSE}';
  frame-src https://pay.psp.example;
  object-src 'none';
  base-uri 'self';
  form-action 'self';
  report-uri https://csp.example.com/report;
  report-to csp-endpoint
Reporting-Endpoints: csp-endpoint="https://csp.example.com/report"
```

Notes that matter here:

- Allowing the vendor host does not stop code inside the vendor's own file. What the allowlist does is block the second stage from an unknown domain, and `connect-src` blocks the beacon home.

- Do not add `'strict-dynamic'` to this policy. In browsers that support it, it makes the host allowlist ignored and trusts anything a nonced script injects.

- Start with `Content-Security-Policy-Report-Only` for a week, fix what breaks, then enforce. A spike of reports for a new domain is exactly the early warning this attack would have produced.

- If a vendor only works with `'unsafe-eval'`, keep it off your checkout and account pages entirely.

### 4. Use Subresource Integrity for static files

SRI makes the browser refuse a file whose hash does not match. It works for versioned libraries that never change:

```
# Generate the hash of the exact file you tested
openssl dgst -sha384 -binary widget-2.4.1.min.js | openssl base64 -A
```

```

```

SRI cannot protect a snippet like Optimizely's, which changes every time someone edits an experiment. That is why it pairs with CSP rather than replacing it.

### 5. Self-host what you can, but know the limit

Hosting a pinned copy of a library yourself removes a third party from the chain. Proxying a dynamic vendor snippet does not: sites that served Optimizely from their own domain still got the malware, because the content came from the compromised project.

### 6. Monitor scripts for changes

Fetch the scripts your pages load on a schedule, hash them and alert on any change, new host or decoding call. A minimal starting point:

```
// check-scripts.mjs - run from cron or CI every few minutes (Node 18+)
import { createHash } from 'node:crypto';
import { readFile, writeFile } from 'node:fs/promises';

const SCRIPTS = ['https://cdn.vendor.example/js/1234567890.js'];
const ALLOWED_HOSTS = ['vendor.example', 'www.yoursite.com'];
const STATE_FILE = './script-hashes.json';

const state = JSON.parse(await readFile(STATE_FILE, 'utf8').catch(() => '{}'));
const alerts = [];

for (const url of SCRIPTS) {
  const res = await fetch(url, { headers: { 'cache-control': 'no-cache' } });
  const body = await res.text();
  const hash = createHash('sha256').update(body).digest('hex');

  if (state[url] && state[url] !== hash) alerts.push(`CHANGED: ${url}`);
  state[url] = hash;

  const hosts = new Set(
    [...body.matchAll(/https?:\/\/([a-z0-9.-]+)/gi)].map((m) => m[1].toLowerCase())
  );
  for (const h of hosts) {
    const ok = ALLOWED_HOSTS.some((a) => h === a || h.endsWith('.' + a));
    if (!ok) alerts.push(`UNKNOWN HOST ${h} in ${url}`);
  }
  if (/\batob\s*\(|\beval\s*\(|new\s+Function\s*\(/.test(body)) {
    alerts.push(`SUSPICIOUS CALL (atob/eval/Function) in ${url}`);
  }
}

await writeFile(STATE_FILE, JSON.stringify(state, null, 2));
if (alerts.length) {
  console.error(alerts.join('\n'));
  process.exit(1); // let cron or CI turn this into an alert
}
console.log('No changes detected');
```

Expect some noise from legitimate vendor updates. Commercial client-side monitoring tools do this inside real browsers, which also catches scripts injected at runtime.

### 7. Isolate the payment page

The skimmer could not read cards typed into a payment provider's cross-origin iframe. Use hosted fields or a hosted payment page, and strip marketing, testing and chat tags from checkout. That single step would have protected card data on many of the affected sites.

![Person holding a payment card while typing on a laptop at a white desk](https://amvprosox.com/api/img/1794920311664-ro8oxc)
*Card details typed into a site's own form were exposed; cards entered in a provider's hosted iframe were not. Photo: Shixart1985, CC BY 2.0, via Wikimedia Commons*

## What PCI DSS 4.0 requires for payment page scripts

| Requirement | What it asks for | Practical control |
| --- | --- | --- |
| 6.4.3 | Every script on a payment page is authorized, its integrity is assured, and an inventory with a written justification exists | Script inventory, approval workflow, CSP allowlist, SRI where possible |
| 11.6.1 | A change and tamper detection mechanism alerts staff to unauthorized changes to security headers and script contents of payment pages as the browser receives them, at least weekly | Script monitoring, CSP reporting, header checks |

Both became mandatory on March 31, 2025. Version 4.0.1 clarified that merchants answer for scripts on their own page, while scripts inside a provider's payment iframe are the provider's job. If you file SAQ A, confirm with your acquirer how the script criteria apply to you.

## What should visitors do?

- Update iPhones and iPads to iOS 18.7.3 or later, or iOS 26.3 or later.

- If you sent crypto on November 16 using an address copied from one of these sites, check where it went, and verify addresses on your hardware wallet's screen.

- If you typed card details into one of these sites that day, watch your statements or ask for a new card.

## The lesson for every site owner

Helme summed it up well: "trusted JavaScript doesn't always remain trustworthy." Your site is only as safe as the weakest dashboard that can push code to it. If you are not sure what runs on your pages, a [website audit](https://amvprosox.com/website-audit) is a good starting point, and ongoing [website maintenance](https://amvprosox.com/website-maintenance) keeps CSP, script inventories and monitoring from drifting.

## Sources

- [Report URI: World's largest brands infected with malware via Optimizely (with November 17 update)](https://blog.report-uri.com/worlds-largest-brands-infected-with-malware-via-optimizely/)

- [Optimizely Status: Web Experimentation Security Incident](https://status.optimizely.com/)

- [StepSecurity: OPTIMJACK, Optimizely-hosted scripts loaded a crypto clipper and card skimmer](https://www.stepsecurity.io/blog/optimjack-optimizely-hosted-scripts-crypto-clipper-card-skimmer)

- [The Hacker News: Hijacked Optimizely scripts served card skimmer and iOS exploit kits](https://thehackernews.com/2026/11/hijacked-optimizely-scripts-served-card.html)

- [BleepingComputer: Optimizely hack infected CNN, Fox News, BBC, Salesforce sites](https://www.bleepingcomputer.com/news/security/optimizely-supply-chain-attack-delivers-malware-to-cnn-fox-news-bbc-salesforce-sites/)

- [Help Net Security: Malware spread through Optimizely hits 178 organizations](https://www.helpnetsecurity.com/2026/11/17/optimizely-vulnerability-malicious-javascript/)

- [Reuters via The Star: Popular websites subverted in bid to steal crypto, target iPhone users](https://www.thestar.com.my/tech/tech-news/2026/11/17/popular-websites-subverted-in-bid-to-steal-crypto-target-iphone-users)

- [Hudson Rock (Infostealers): Beyond Optimizely, malicious scripts traced across four web services](https://www.infostealers.com/article/beyond-optimizely-malicious-scripts-traced-across-four-web-services/)

- [SecurityWeek: Optimizely vulnerability exploited to serve malware on major websites](https://www.securityweek.com/optimizely-vulnerability-exploited-to-serve-malware-on-major-websites/)

- [Reuters via Honolulu Star-Advertiser: Popular websites served malicious code targeting crypto, iPhones](https://www.staradvertiser.com/2026/11/17/breaking-news/popular-websites-served-malicious-code-targeting-crypto-iphones/)

- [Newsweek: Cyber attack hits 178 major websites, full list of organizations](https://www.newsweek.com/cyber-attack-websites-affected-12725863)

- [36Crypto: Ledger confirms website security breach linked to Optimizely cyberattack](https://36crypto.com/ledger-confirms-website-security-breach-linked-to-optimizely-cyberattack/)

- [Gizmodo: Sprawling malware attack reportedly affects Zoom and other major brands](https://gizmodo.com/sprawling-malware-attack-reportedly-affects-zoom-cnn-fox-news-peacock-and-170-others-2000840831)

- [PCI Security Standards Council: PCI DSS document library](https://www.pcisecuritystandards.org/document_library/)

## FAQ

### What is the cyber attack today on CNN, BBC and Zoom?

It is the Optimizely supply-chain attack of November 16, 2026. A flaw in Optimizely Web Experimentation let attackers add malicious JavaScript that ran on sites of at least 178 organizations.

### Was my iPhone hacked by the Optimizely script?

Researchers confirmed the Coruna and DarkSword exploit pages were delivered to older iOS versions. As of November 18, no successful compromise of a real device has been confirmed, and Hudson Rock says its evidence does not show one. Updating to iOS 18.7.3 or later, or iOS 26.3 or later, closes the flaws these kits target.

### Is the Optimizely attack over?

The active injection is over. Report URI says every infected file was clean by 23:11 UTC on November 16, and the payload and injector domains went offline early on November 17. Optimizely's incident was still marked as monitoring on November 18, and the attacker moved the stolen crypto that morning.

### Would a Content Security Policy have blocked the attack?

Mostly yes, if it used a host allowlist in script-src, a tight connect-src, and no unsafe-eval or strict-dynamic. Report URI found that the 24 organizations with such a policy blocked the payload.

### Were other services besides Optimizely used in the attack?

Yes. Hudson Rock traced the same loader infrastructure through Kissmetrics, StorifyMe and the minute.ly video player, starting November 13. How those services were abused is still unknown, and no attacker has been named.

### What should Optimizely customers do now?

Review Project JavaScript and experiment code, audit users and API tokens, rotate secrets in non-HttpOnly cookies, treat cards entered on your own forms on November 16 as exposed, and ask Optimizely which projects changed.
