AI Engine EUVD-2026-36916

Exploitation requires an authenticated WordPress session with at least the Editor role on a site running AI Engine ≤ 3.4.9; no user interaction from another user is needed (UI:N) and the request can be delivered remotely over the network (AV:N) to the WordPress admin or REST endpoints. Limiting factors: the attacker must possess or compromise a valid Editor (or higher non-admin) account, so sites that issue Editor privileges sparingly, enforce strong authentication on those accounts, or restrict /wp-admin access by IP face significantly reduced risk. Sites without the AI Engine plugin installed, or already on a release post-3.4.9, are not affected.

The CVSS 3.1 vector (AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H) reflects a network-reachable but high-privilege-required flaw with full CIA impact once exploited - appropriate for a WordPress plugin endpoint that already requires an authenticated Editor session. EPSS and KEV signals are not provided in the input; absent KEV listing and absent any known public POC, real-world urgency is moderate rather than emergency-grade. The realistic threat model is insider abuse or a compromised Editor account being used as a stepping stone to full site takeover, which on multi-author WordPress deployments (publishers, agencies, membership sites) is a meaningful escalation path. Note: CVSS labels PR as High, but the WordPress role model typically treats Editor as a mid-tier role - the rating likely reflects that any non-anonymous, trusted-author account is required, which defenders should weigh against how freely Editor accounts are issued in their environment.

An attacker who already controls a WordPress Editor account - either an insider, a compromised contributor on a multi-author blog, or an account taken over via phishing or password reuse - logs into the site and invokes a vulnerable AI Engine endpoint that fails to properly enforce the Administrator capability. The endpoint promotes the attacker's user to Administrator or otherwise grants administrative capability, after which the attacker installs a malicious plugin, modifies theme files for persistence, and exfiltrates site data. No public exploit identified at time of analysis, but the network attack vector and low complexity make weaponization straightforward once details are reverse-engineered from the patch.