Key details

  1. Incident occurred 16 November 2026; Optimizely published a 17 November incident report.
  2. Vendor says unauthorised user impersonation, not stolen customer passwords, enabled Project JavaScript modification.
  3. Vendor reports vulnerability closed by 16:10 UTC and malicious code removed by 22:24 UTC on 16 November.
  4. Report URI found 188 malicious snippets among 1,611 examined, affecting 178 organisations.
  5. Only Optimizely Web Experimentation is confirmed affected; SDKs and Feature Experimentation were not.
  6. Payload attempted payment-card and cryptocurrency theft; actual successful harm remains uncertain.
  7. Optimizely says hosted payment iframes were not accessible to the malicious script.

What builders should take away

  1. Ask Optimizely for confirmation whether any of your Web Experimentation projects were affected.
  2. Audit Project JavaScript and changes around 16 November UTC and preserve relevant browser/site logs.
  3. Review exposure of page-owned payment fields, non-HttpOnly cookies and sensitive tokens; rotate or invalidate where warranted by evidence.
  4. Review enforce-mode CSP and connect-src restrictions; a vendor allowlist alone cannot guarantee safety.
  5. Monitor and inventory third-party scripts and restrict who can change experimentation code.

What changed

On 16 November 2026 an unauthorised party exploited an Optimizely Web Experimentation flaw that allowed impersonating customer users and editing Project JavaScript. Optimizely's incident report says the first malicious code was written at 07:51 UTC, the company detected activity at 13:34, closed the vulnerability at 16:10 and removed injected code from affected projects by 22:24. Report URI researcher Scott Helme independently found 188 malicious snippets associated with 178 organisations among 1,611 checked. The payload attempted cryptocurrency address substitution, payment-card skimming and other browser attacks. Optimizely says only Web Experimentation was affected, not Feature Experimentation or its SDKs.

Why it matters

A trusted experimentation vendor's script can execute on many customer sites without each site being individually breached. The malicious code was inserted into a customer-controlled Project JavaScript field and served by the vendor's legitimate delivery infrastructure. The incident demonstrates why a permitted third-party CDN is not equivalent to trusted code, and why CSP design and vendor-script change monitoring matter. Reuters could not establish how many users were harmed; observed payload capabilities must not be described as confirmed successful theft from all affected visitors.

The attack crossed a vendor authorisation boundary

Optimizely says the attacker did not steal customer passwords or compromise an administrator account. Instead, an authorisation flaw let an unauthorised party impersonate users and change some customers' Project JavaScript. The injected code then ran in visitors' browsers where the Web Experimentation snippet was loaded. This is distinct from a compromise of each customer's own server.

Scope and timeline are partly independently measured

Optimizely reports the first malicious write at 07:51 UTC on 16 November, detection at 13:34, vulnerability closure at 16:10, and full removal by 22:24. Report URI found 188 affected snippets across 178 organisations in its collected sample. The vendor has not confirmed the same count, and independent researchers may not have captured every affected snippet.

Malware sought payments and session information

Independent analysis found code designed to replace cryptocurrency addresses, capture card details typed into page-owned forms, and steal readable browser data. Reuters also reports attempts to deliver exploit code to older iPhones. Optimizely characterises the intended harm as payment and cryptocurrency theft and says payment details entered into separately hosted provider iframes were not accessible to the script. Neither these capabilities nor the presence of malicious code establishes a successful compromise of every visitor.

Response requires browser-side exposure assessment

Optimizely says it fixed the vulnerability, removed injected code and contacted affected customers. Operators should obtain direct confirmation of whether their projects were affected, examine Project JavaScript and change history, correlate site traffic during the exposure window, review possible payment/session exposure, and consider credential or session remediation where evidence warrants. Review CSP enforcement and third-party script monitoring rather than assuming an allowlisted vendor host is safe.

What to watch next

  • Verified affected-customer scope and actual user losses.
  • Additional technical disclosure of the authorisation vulnerability and independent postmortems.
  • Further attacker infrastructure, indicators and remediation guidance.
  • Any evidence of exposure beyond Web Experimentation.

Still unclear

  • Report URI's affected-organisation count is a sampled independent measurement, not an Optimizely-confirmed total.
  • Reuters could not verify successful theft or total user impact.
  • The exact vulnerability mechanics and all exploit prerequisites are not fully public.

Sources

Direct reading behind this dossier.

3 sources

Discussion

Discussion is reader-contributed. Comments are not part of the BTN dossier or its editorial evidence.

0 visible comments

Join the discussion

Keep comments useful and relevant. Reader contributions may be moderated and are not BTN editorial evidence.

Sign in to comment