How the Optimizely Supply-Chain Attack Works
Modern business websites rarely run code from a single server. A typical corporate site loads analytics tags, A/B testing scripts (like Optimizely), chat widgets, and marketing pixels directly from external vendors. Every time a visitor loads a page, their browser fetches and executes JavaScript from those third-party domains—granting that external code the exact same permissions inside the browser tab as the company's own code.
In the incident disclosed on November 16, 2026, attackers did not break into the web servers of Zoom, BBC, HSBC, or IKEA. Instead, they compromised Optimizely customer projects and injected malicious JavaScript into the platform's script delivery pipeline. Because browsers implicitly trusted the Optimizely tag on those sites, visitors automatically ran the attacker's code.
Live Data-Flow Diagram
Toggle between an unprotected browser session and a CSP-hardened website.
UNPROTECTED: Malicious JS runs in first-party DOM and exfiltrates card data, cookies, and swapped crypto addresses.
Malicious code is planted inside Optimizely customer projects
Using an issue in Optimizely's account/project workflow, attackers inserted custom JavaScript payloads directly into customer experimentation projects—affecting 188 snippets across 178 organizations out of 1,611 examined by Report URI.
Visitors automatically fetch the script from an allowlisted CDN
Because organizations intentionally embed Optimizely snippets in their HTML <head>, every visitor's browser fetches the script automatically over HTTPS from a legitimate third-party resource—even when the company's own server is completely uncompromised.
Code executes inside the website's live browser session
Unless sandboxed in a cross-origin iframe or blocked by Content Security Policy, external JavaScript inherits full read/write access to the page DOM, form inputs, clipboard events, and non-HttpOnly cookies.
Multi-vector harvesting runs silently in the background
The attacker payload targets high-value outcomes simultaneously: swapping cryptocurrency addresses, skimming payment cards, capturing cookies and screenshots, and probing Apple iOS visitors for exploit kit delivery.
Inside the 4-in-1 Malicious Payload
Rather than relying on a single theft mechanism, the injected Optimizely snippets delivered a multi-purpose payload engineered to monetize every type of visitor. Hover or tap each card below to flip and inspect how it operated.
Cryptocurrency Theft (Crypto-Clipping)
Detects wallet addresses copied or displayed on screen and replaces them with attacker-controlled wallets.
How Crypto-Clipping Works
The script intercepts clipboard copy/paste events and scans DOM text for Bitcoin, Ethereum/EVM, TRON, and Solana wallet patterns. When a user copies an address to send funds, the clipboard is silently overwritten—turning an irreversible blockchain transfer into an instant loss. Defense: Always verify wallet addresses character-by-character out-of-band before signing.
Payment Card Skimming (Magecart-Style)
Hooks checkout forms and keyboard events to steal card numbers, expiry dates, and CVVs before submission.
How Browser Skimming Works
By attaching event listeners to input fields or wrapping fetch()/XMLHttpRequest, the script captures cardholder data as it is typed and exfiltrates it alongside customer billing details. Defense: Host payment fields inside cross-origin PCI-compliant iframes (Stripe Elements, Adyen Drop-in) where parent-page JS cannot read inputs.
Cookie & Screenshot Exfiltration
Reads accessible browser cookies and captures visual snapshots of authenticated user sessions.
How Session Harvesting Works
Any cookie lacking the HttpOnly flag can be read directly via document.cookie, exposing session tokens and PII. Combined with DOM/canvas screenshot capture, attackers obtain confidential account screens. Defense: Enforce HttpOnly; Secure; SameSite on every auth cookie and block unknown outbound endpoints via CSP connect-src.
iOS Exploit Kit Delivery
Fingerprints Apple iOS browsers and attempts to serve exploit kits (Coruna / DarkSword) to unpatched devices.
How Mobile Kit Delivery Works
Technical analyses from Report URI and SOCRadar noted conditional logic checking for iOS devices and attempting to stage browser/WebKit exploit chains depending on OS version. Defense: Keep iOS, Safari, Chrome, and Edge patched immediately on all corporate and BYOD devices, and enable Lockdown Mode for high-risk executives.
Affected Sectors & Notable Organizations
Out of 1,611 Optimizely snippets examined, investigators identified 178 affected organizations spanning enterprise technology, global media, banking, retail, airlines, telecom, and cryptocurrency—including widely visited Canadian and North American services such as CTV, Starbucks Canada, Zoom, GitLab, Salesforce, and Equifax.
Technology
- Zoom
- GitLab
- Salesforce
- Bitly
- Hootsuite
News & Media
- BBC
- CNN
- Fox News
- NPR
- CTV (Canada)
Banking & Finance
- HSBC
- Equifax
- ABN AMRO
Retail
- IKEA
- ASOS
- Levi's
- American Eagle
- Crocs
Airlines & Travel
- Air France-KLM
- Alaska Airlines
- Transavia
Cryptocurrency
- Ledger
Telecommunications
- Vodafone Australia
- Virgin Media
- O2
Other Notable
- Starbucks Canada
- Change.org
- Honeywell Home
Source: Scott Helme / Report URI & BleepingComputer public investigation disclosures (Nov 16, 2026). Almost all malicious snippets were removed by 21:52 UTC Nov 16; listing does not imply active infection.
Why Firewalls Missed It — And How Strict CSP Stops Exfiltration
Because a third-party script is requested directly from the visitor's browser to the vendor CDN, perimeter firewalls and server WAFs never inspect the malicious script. Even worse, once the script runs, many sites have overly broad Content-Security-Policy headers—or none at all—allowing the malware to send stolen credit cards and cookies to any attacker server on the internet.
Content-Security-Policy Comparison
Content-Security-Policy: default-src * 'unsafe-inline' 'unsafe-eval'; # ❌ Wildcard (*) allows JS to load and exfiltrate stolen cards/cookies to ANY C2 domain # ❌ No connect-src restriction, no report-to telemetry, payment inputs in main DOM
Content-Security-Policy: default-src 'self'; script-src 'self' https://cdn.optimizely.com https://www.googletagmanager.com; connect-src 'self' https://logx.optimizely.com https://www.google-analytics.com; frame-src 'self' https://js.stripe.com; report-to csp-endpoint;
1. Lock Down connect-src
Even if a trusted vendor script is hijacked, strict connect-src, form-action, and img-src rules block exfiltration to attacker C2 servers.
2. Isolate Checkout Pages
Never run A/B testing, chat, or marketing tags inside payment, login, or MFA pages—and keep card fields inside PCI cross-origin iframes.
3. Monitor CSP Reports
Point report-to telemetry directly to your SIEM/SOC so unauthorized outbound domains trigger an immediate alert.
30-Second Third-Party Script Risk Check
Answer 4 quick questions about your company's websites, portals, or e-commerce stores to gauge your exposure to supply-chain script attacks like the Optimizely incident.
1.Do your websites load third-party scripts (Optimizely, analytics, chat widgets, pixels, or Tag Manager)?
If your site loads external vendor JavaScript directly in the browser.
2.Do any of those third-party tags run on login, banking, portal, or checkout pages?
Sensitive forms without iframe isolation expose inputs and session state.
3.Does your site lack a strict Content-Security-Policy restricting connect-src & script-src?
Without CSP, a hijacked script can send data to any server on the internet.
4.Are session cookies missing HttpOnly flags, or do vendor portals lack SSO/MFA?
Cookies without HttpOnly can be read by malicious JS; weak vendor logins enable project hijacking.
Required Precautions & Action Checklists
How QueryTel Protects Canadian Organizations
Supply-chain incidents happen outside your firewall. QueryTel combines application hardening, browser security policies, and 24/7 SOC monitoring so a compromised vendor doesn't compromise your customers or staff.
Third-Party JS & CSP Hardening
We audit every external script, Tag Manager container, and HTTP security header on your web properties—locking down script-src and connect-src without breaking marketing workflows.
Explore service24/7 SOC 24 & DNS Monitoring
Our Canadian SOC monitors endpoint telemetry, DNS queries, and CSP violation logs around the clock to catch unauthorized C2 beacons or browser exploit activity early.
Explore serviceZero-Trust Endpoint & Browser Patching
Automated patching across Windows, macOS, iOS, and browsers ensures mobile exploit kits like those seen in this incident fail against updated devices.
Explore serviceManaged IT & Vendor Governance
Full lifecycle management, SaaS SSO/MFA enforcement, and incident response support for Canadian businesses and multi-site operations.
Explore serviceIncident Timeline & Verified Sources
NOVEMBER 16, 2026 — DISCOVERY
Cybersecurity researcher Scott Helme (Report URI) discovers malicious JavaScript injected into Optimizely customer projects, impacting major global brands across 1,611 examined snippets.
NOVEMBER 16, 2026 — ECOSYSTEM IMPACT MAPPED
188 malicious JS snippets confirmed across 178 organizations—delivering crypto-clipping, card skimmers, cookie/screenshot exfiltration, and iOS exploit kits.
NOVEMBER 16, 2026 · 21:52 UTC — REMEDIATION
Original researcher confirms almost all malicious scripts have been removed. Precautionary monitoring, CSP enforcement, and third-party script audits remain essential.
Frequently Asked Questions
Were the affected organizations’ own web servers compromised?
No. In a third-party JavaScript supply-chain attack, the malicious code is delivered through an external vendor script embedded on the page. An organization’s own servers can be completely patched and unbreached while still serving the compromised script to visitors.
Are all listed websites still infected today?
No. Scott Helme reported that almost all malicious scripts were removed by 21:52 UTC on November 16, 2026. The advisory and list of organizations are precautionary and do not mean all listed websites are currently compromised.
Should employees change passwords if they visited one of these sites?
If an employee entered corporate credentials or payment details on an affected checkout or login flow during the November 16 exposure window, rotate those credentials immediately and notify IT so session tokens can be revoked and monitored.
What is the single highest-impact fix for our own company website?
Combine strict Content-Security-Policy (CSP) headers—specifically locking down script-src and connect-src—with isolating sensitive login and payment pages from marketing or A/B testing tags. Also enforce HttpOnly, Secure, and SameSite flags on all session cookies.
Not sure what scripts are running on your website?
QueryTel's security team will inspect your external JavaScript dependencies, HTTP security headers, and Content Security Policy—and show you exactly how to close supply-chain risk.
Was this advisory helpful?
Help us keep our threat advisories accurate and actionable for Canadian teams.
Related Security Advisories
Cisco Firewall Zero-Day (CVE-2026-76418)
Actively exploited SSL VPN crash bug on ASA & FTD.
Web AppSecThe Silent Web Root Liability
Hardening enterprise PHP applications, headers & forms.
SOC 24How AI Works Inside a Modern SOC
How analysts and AI triage thousands of alerts in real time.