OneDot61 works with ImmuniWeb on a commercial basis.
Report URI found that on 16th November 2026, malicious code was served to visitors of 178 organisations. These included HSBC Australia, Vodafone Australia, TPG Telecom, Rest Super, Zoom and the BBC. It was served through a script those sites trusted: Optimizely's. None of the websites changed their own code. Here is a brief summary of what happened and what helped.
Compromised Third-Party Scripts from Suppliers
Many websites use a supplier's script to test or personalise their pages. The website owner adds a line to each page that loads the script, and the supplier hosts that script on its own systems, where it can update it at any time. When someone visits the page, their browser fetches the script from the supplier and runs it. If an attacker gets into the supplier's systems and changes the script, every visitor's browser runs the attacker's version, even though the website owner changed nothing.
The website owner did nothing wrong, and the supplier may not know its script was compromised, but website visitors are exposed. The ASD's guidance on managing cyber supply chains explains how to set cyber security expectations with your suppliers.
The Optimizely Incident in Brief
Report URI's write-up has the full details. In brief, Report URI found malicious JavaScript being served to visitors of 178 organisations. It checked 1,611 snippets and found the code in 188 of them, including those used by HSBC Australia, Vodafone Australia, TPG Telecom, Rest Super, Zoom, the BBC, CNN, Salesforce and GitLab.
- What the added script did: it swapped cryptocurrency wallet addresses, skimmed payment card details typed into a site's own form fields, collected cookies and screenshots, and on iPhones and iPads delivered the Coruna and DarkSword exploit kits.
- Timing (AEDT): Report URI first saw the injection at 3:21 pm on Monday 16th November. By 8:52 am on Tuesday 17th November it found the code in only one of the 188 snippets.
- Other suppliers: the same attacker also used other suppliers' scripts. iTWire reports poisoned Kissmetrics and StorifyMe scripts were observed from 13th November, including in a 14th November copy of the Sydney Morning Herald's website. StorifyMe confirmed that unauthorised JavaScript reached some of its customers' web integrations.
Optimizely's status page, which still listed the incident as monitoring on 18th November, says:
"...an unauthorized party exploited a vulnerability in Experimentation and used it to add malicious JavaScript to some customer projects. We've fixed the vulnerability and removed the injected code from affected projects. We've also taken steps to secure affected user accounts and put additional protections in place to prevent unauthorized code from being delivered."
According to iTWire, Optimizely has said the vulnerability was in its Experimentation account signup flow. They have not explained how that gave the attacker access to customer projects.
A Content Security Policy Helped Mitigate the Attack
A Content Security Policy (CSP) is a list of rules the website owner writes, naming the domains its pages may load scripts from and send data to. The website sends this list with every page, and the visitor's browser enforces it. In this attack, the supplier's domain was already on the list, so the browser loaded the changed script as normal. But when that script tried to load more code from the attacker's domains, or send stolen data to them, the browser checked the list, found those domains weren't on it, and blocked the request. Nothing had changed on the website's own servers, so the CSP running in the browser was the one control still in a position to stop the attack.
Report URI assessed the CSP of 159 affected organisations:
- At 68, there was no CSP.
- At 50, the payload ran and could send data out.
- At 17, the payload ran, but the CSP stopped it sending data to the attacker.
- At 24 (15%), the CSP blocked the payload.
BleepingComputer reports that one version of the code could still run even if the main malicious script was blocked, so blocking scripts alone isn't enough. At the 17 sites where the payload ran, the CSP still stopped it sending data to the attacker's domains, so visitors' data didn't reach the attacker.
Strict CSPs are uncommon because many marketing and testing tools need permissions that weaken them, which helps explain why only 24 of the 159 organisations blocked the payload. A CSP can also report unexpected scripts and connections, which gives early warning when something on a site changes.
A Free Website Security Test to Start With
ImmuniWeb's Website Security Test is a free online tool with a PDF report, and it includes a Content Security Policy test. It won't prevent what happened with Optimizely, but it's a quick way to check the basic security on your own website. ImmuniWeb describes it as a way to "test your website and web server for security, privacy, encryption, protection from data scraping, and compliance with GDPR and PCI DSS", with:
- Web Security Vulnerabilities Scan
- Content Security Policy (CSP) Test
- AI Bot Protection Test
- HTTP Security & Privacy Headers Test
- DNSSEC Configuration Test
- GDPR & PCI DSS Compliance Test
As of 17th November 2026, ImmuniWeb reports more than 199 million tests performed. ImmuniWeb is available through OneDot61.
Related Reading
- ASD's 2025-26 Cyber Threat Report: 82k Cybercrimes Reported and 5 Actions to Take
- Detecting Agentic AI Attacks Inside Your Network
Explore Our Portfolio
Need cyber security tools? Our solutions span AI security, AI governance, cloud, zero trust, email security, network detection and response, endpoints, pen testing, post-quantum encryption and more. Contact OneDot61.