Contact Zero

CVE-2026-97670

The Avada (Fusion) Builder plugin for WordPress is vulnerable to authorization bypass in all versions up to, and includi

score 35CRITICAL 9.1

Summary

The Avada (Fusion) Builder plugin for WordPress is vulnerable to authorization bypass in all versions up to, and including, 7.16.1. This is due to the plugin not properly verifying authorization before dispatching a WordPress action hook whose name is taken from an attacker-supplied form-field value (via the notification email_message [field] placeholder and the {action_hook,...} dynamic-data token; the 3.16.1 trust gate is_content_request_supplied() only inspects $_POST['args']/$_GET['args'], never the $_POST['formData'] the public form-submit endpoint parses). This makes it possible for unauthenticated attackers to invoke arbitrary WordPress action hooks (multiple per request), causing state changes up to permanent, irreversible destruction of site content: a verified unauthenticated request permanently deleted trashed posts, pages, and comments via the core wp_scheduled_delete action. Other non-deny-listed hooks extend the impact to denial of service (e.g. wp_maybe_auto_update) and, where vulnerable third-party handlers are installed, further privileged writes. The same unauthenticated dynamic-data pipeline additionally exposes a blind arbitrary user/post-meta read; the read result is delivered only to the site owner and is not attacker-exfiltrable through the plugin's own email/response paths. Exploitation requires a published Avada form with AJAX submission and a notification whose email_message template includes an [all_fields] or explicit [field] placeholder - the default form configuration.

Published 2026-10-10 · first seen here 2026-10-10

Analysis & write-ups

No technical write-up from a research team yet. Contact Zero checks Unit 42, Google/Mandiant, Microsoft, Talos, CrowdStrike, Rapid7, watchTowr and others several times a day, plus NVD and CISA reference links.

Where it shows in your logs

Server or web application · Command or code injection (remote code execution) medium confidence

What to look for

Data sources

Web server and WAF logs · ATT&CK DS0015 Application Log Content

Sentinel
W3CIISLogAppServiceHTTPLogsAzureDiagnostics (Application Gateway / Front Door WAF)_Im_WebSession
Splunk
Web.Webms:iis:autoaccess_combinedWAF / proxy sourcetypes
CrowdStrike
Falcon Next-Gen SIEM: third-party web / WAF / proxy logs

Process creation · ATT&CK DS0009 Process Creation

Sentinel
DeviceProcessEventsSecurityEvent (4688)Sysmon Event ID 1_Im_ProcessCreate
Splunk
Endpoint.ProcessesXmlWinEventLog:Microsoft-Windows-Sysmon/OperationalWinEventLog:Security (4688)
CrowdStrike
ProcessRollup2SyntheticProcessRollup2

File creation · ATT&CK DS0022 File Creation

Sentinel
DeviceFileEventsSysmon Event ID 11_Im_FileEvent
Splunk
Endpoint.FilesystemSysmon Event ID 11
CrowdStrike
NewExecutableWrittenNewScriptWritten*FileWritten events

Network connections · ATT&CK DS0029 Network Connection Creation / Network Traffic Flow

Sentinel
DeviceNetworkEventsCommonSecurityLog (firewall)_Im_NetworkSession
Splunk
Network_Traffic.All_TrafficFirewall sourcetypes (pan:traffic, fortigate_traffic, cisco:asa)
CrowdStrike
NetworkConnectIP4NetworkReceiveAcceptIP4NetworkConnectIP6

A generic baseline worked out from the product type and weakness (CWE-94), not a detection. Check table and field names against your environment. Hunt queries for Sigma, Splunk, Sentinel and CrowdStrike are coming.

References