What the Optimizely incident means for A/B testing security

·Updated November 2026
Key Takeaways
  • We have no indication that Varify or Varify customers were affected. Right after the incident became known, we also scanned the JavaScript code stored in Varify accounts. We found no incidents and no dangerous entries.
  • Every testing tool that lets you add custom JavaScript carries this kind of risk. What matters is who can change that code and how fast you notice.
  • A Content Security Policy (CSP) is the strongest protection on your own side.

On November 16, 2026, visitors of many large websites received malicious JavaScript through their A/B testing tool. Since then, many of you have asked us the same question: could this happen with Varify? Here is a straight answer, what we do to protect your website, and what you can check yourself in about 15 minutes.

What happened

The security company Report URI found malicious JavaScript served to visitors of 178 organizations on November 16, all through Optimizely Web. The attacker had written code into the Project JavaScript field of customer projects, a field that runs on every page. The rest of each site’s Optimizely snippet was unchanged and still loaded from Optimizely’s own CDN, so the code came from a trusted domain. According to Optimizely’s initial report, an unauthorized party exploited a vulnerability in Optimizely’s sign-up process and inserted malicious JavaScript into some customer projects. The code swapped cryptocurrency wallet addresses and skimmed payment card details and cookies.

The full incident report was published by Report URI.

Why this concerns every testing tool

An A/B testing tool runs on your pages by design. It also lets you add your own code, for example to change a page or to track a goal. If someone gets write access to that code, it runs for every visitor, and it comes from a source your website trusts.

That is why we look at two questions: who can change the code, and how fast would you see a change?

What we did when the incident became known

Right after the Optimizely incident became known, we scanned the JavaScript code stored in Varify accounts for malicious entries. We found no incidents and no dangerous entries.

How Varify protects your account

Who can change the code

  • Roles and permissions limit what each team member can do.
  • MFA protects every login, and the account owner can make it mandatory for the whole team.
  • Login verification asks for an email confirmation when someone logs in from a device, IP address or location Varify does not recognize. A stolen password alone is not enough to get in.
  • Single sign-on (SSO) is available.
  • Sensitive action confirmation can be switched on, so that users must enter their password again before they change an experiment, Experiment JavaScript, Global JavaScript or Global Tracking JavaScript. It is off by default.
  • Cloudflare Turnstile protects the login against automated attacks.

How fast you see a change

  • The activity log records every change with who, what and when, including the JavaScript and CSS of variations, Experiment JavaScript and CSS, Global JavaScript and Global Tracking JavaScript.
  • The login history and the active sessions show who logged in from where.

What protects you at platform level

  • Strict tenant separation. Every Varify account is strictly separated from all others. It is not possible to write into one account from another, and no endpoint allows it, including the sign-up.
  • Restricted staff access. Access by Varify staff to customer accounts is restricted and logged. Staff reach our systems only through an extra layer of protection that requires two-factor authentication.
  • Emergency stop. We can stop the delivery of Varify immediately with rules at our network provider, without any change on your website. You can also remove the snippet or deactivate it in your tag manager at any time.
  • Notification. If a security incident affects your data, we notify you immediately after becoming aware of it, and within 48 hours at the latest.

What you can do in 15 minutes

You do not have to do anything. But you can:

  1. Open the activity log and look for changes to Global JavaScript, Global Tracking JavaScript and experiment code that nobody on your team made.
  2. Read your Global JavaScript and Global Tracking JavaScript for code you do not know, especially encoded strings or unfamiliar domains.
  3. Check the login history and active sessions, and review team members and API keys.
  4. Make MFA mandatory and activate the sensitive action confirmation.
  5. Use a Content Security Policy.
  6. Add a tool that watches your scripts, and set up notifications.

Why a Content Security Policy matters most

In the Optimizely incident, the injected code loaded a script from an attacker’s domain. A CSP that allows only the domains you trust stops the browser from loading such a script, and it can limit where scripts can send requests. Report URI found that a restrictive script-src made the difference, and that allowing 'unsafe-eval' let the attack through in some cases. Varify works with a strict policy and does not need unsafe-eval. You find the domains Varify uses in our CSP documentation.

One limit you should know: Varify can only check JavaScript that is added directly in your account. Varify has no access to external scripts that your code loads from other domains. The owner of such a script can change it at any time, and nobody would notice. A CSP limits which external domains can run on your website at all.

For ongoing protection, add a tool that watches the scripts on your website, such as Script Watch. Report URI offers tools like Script Watch to detect changes to scripts that you trust. Set up notifications by email or Slack, if your tool supports it, so your team sees a change when it happens.

Questions?

If you want help reviewing your setup or your CSP, write to us at security@varify.io. More details are in our security features overview and in the article on login verification.