AI Engine CVE-2026-110982

Exploitation requires an authenticated account with the WordPress Editor role or higher on a site running AI Engine prior to 3.8.6; the vulnerable code path updates an object's title and excerpt without first checking the caller's edit capability for that specific object. There is no special plugin configuration or feature toggle named in the description - the missing-authorization defect is present in the default plugin behavior. Limiting factors: the attacker must already possess Editor-level (or higher) privileges (PR:H), so untrusted or unauthenticated outsiders cannot exploit it directly; impact is bounded to the title and excerpt fields of targeted objects rather than full content control; and a vendor fix in 3.8.6 removes the condition entirely.

This is a genuine but moderate-priority issue, not a critical one. The provided CVSS 3.1 vector (AV:N/AC:L/PR:H/UI:N/S:U/C:N/I:H/A:N, base 4.9) correctly characterizes it as a network-reachable, low-complexity flaw that nonetheless requires high privileges (PR:H) - specifically an authenticated WordPress user already holding the Editor role or above. The signals are internally consistent: CWE-862 (Missing Authorization) matches the description of a broken access-control check rather than remote unauthenticated compromise. A public exploit exists (WPScan), which raises exploitability, but there is NO CISA KEV listing and NO EPSS score was provided, so there is no evidence of active, widespread exploitation - POC availability and active exploitation are distinct signals and must not be conflated. The real limiting factor is the PR:H precondition: an attacker must already be a trusted Editor+, so this is primarily an insider / privilege-escalation-within-the-CMS concern (an Editor altering content they shouldn't, e.g. navigation menus, block templates, global styles, WooCommerce coupons) rather than an externally exploitable breach. One caveat to verify: the vendor rates integrity as I:H, but the actual capability is limited to overwriting two fields (title and excerpt), which may overstate impact for many object types - see assessed vector below.

A low-trust or compromised WordPress Editor account on a site running AI Engine before 3.8.6 sends a crafted request to the plugin's object-update endpoint, which fails to verify the caller may edit the targeted object. The Editor overwrites the title and excerpt of objects they normally cannot touch - navigation menus, reusable block templates, global styles, or WooCommerce coupons - to deface content, mislead users, or disrupt store configuration. A public exploit is available via WPScan, lowering the barrier to abuse.