Your Website Trusts the Script. Should Your Security Team?
A third-party JavaScript integration can become an attack route even when it loads from a legitimate domain. The recent incident involving 178 organisations shows why script permissions and browser security controls deserve closer scrutiny.

On 16 November 2026, malicious JavaScript was delivered to visitors of websites belonging to 178 organisations through Optimizely Web Experimentation. Security researcher Scott Helme identified 188 infected snippets among 1,611 examined. The affected organisations included major names in finance, retail, technology, media and travel. [1]
The attacker inserted malicious code into customer projects within the experimentation platform. The original JavaScript snippet continued to load from Optimizely's legitimate content delivery domain. This meant that a website could receive compromised code through a source it had already approved.
The payload attempted to replace cryptocurrency wallet addresses, capture payment card details entered into website forms, steal certain browser cookies and take screenshots. It also attempted to deliver exploit code targeting Apple mobile devices. Reuters reported that the extent of successful data theft and the number of affected visitors could not be established. [1][2]
The technical issue goes beyond the compromised vendor. It exposes a weakness in how organisations define trust inside their browsers.
A trusted domain is not a guarantee of trusted code.
Many websites use Content Security Policy (CSP) to tell browsers which sources are permitted to load scripts and send data. A typical policy may allow a trusted vendor's domain because the website needs its functionality.
But a source allowlist answers only one question: where may the browser load code from? It does not establish whether the code coming from that source is safe.
In this incident, the malicious loader arrived through the approved Optimizely domain before attempting to load additional scripts and communicate with attacker-controlled destinations. The initial trusted source therefore became the entry point for the next stages of the attack. [1][3]
Report URI examined CSP configurations on affected websites and found a substantial control gap. Of the 159 organisations assessed, 68 had no CSP. Another 50 had policies that allowed both the malicious payload and data theft. Seventeen allowed the payload to run but blocked its outbound data transmission. Only 24 blocked the payload outright. [1]
This distinction matters. Blocking suspicious outbound connections can reduce data exfiltration, but it may not prevent malicious code from changing information displayed on a page. A cryptocurrency address replacement, for example, can happen entirely within the browser.
Organisations should review CSP as an active security control rather than a one-time configuration task. At a minimum, security teams should:
-
Restrict
script-srcto explicitly approved sources and assess whether nonce-based policies are appropriate. -
Review
connect-srcto limit where browser scripts can transmit data. -
Avoid allowing unsafe execution features such as
unsafe-evalunless there is a documented technical requirement. -
Monitor changes to third-party scripts and investigate unexpected external domains.
-
Test payment forms, authentication flows and other sensitive browser functions under realistic attack scenarios.
The review should also cover the full web application attack surface. Teams need to understand which pages expose sensitive functions, how scripts interact with those functions and whether weaknesses in application logic could increase the impact of a compromised integration.
Authorised penetration testing can help identify weaknesses in supported web applications, APIs and exposed access paths. Bronyx AI, ITSEC's AI-assisted autonomous penetration testing platform, can support security assessments within its confirmed technical scope. CSP configuration and browser policy reviews should remain part of the organisation's wider application security programme.
A website's security depends on every component that executes within it. If a third-party script can influence a payment form, user session or browser connection, it deserves the same scrutiny as other parts of the application.
Assess your supported digital assets through authorised testing with Bronyx AI: https://bronyx.ai/
References
-
Report URI, “World's Largest Brands Infected With Malware Via Optimizely,” 16 November 2026. https://blog.report-uri.com/worlds-largest-brands-infected-with-malware-via-optimizely/
-
Reuters, “Popular websites compromised in bid to steal crypto, target iPhone users,” 17 November 2026. https://www.reuters.com/legal/government/popular-websites-subverted-bid-steal-crypto-target-iphone-users-2026-11-17/
-
Optimizely World, “Lessons From the Optimizely Web Experimentation Compromise,” 17 November 2026. https://world.optimizely.com/blogs/mark-stott/dates/2026/11/lessons-from-the-optimizely-web-experimentation-compromise
.png)
.png)

