CVE-2026-16953: 
WordPress Analyse et atténuation des vulnérabilités

CVE-2026-16953 is an Insecure Direct Object Reference (IDOR) / Authorization Bypass vulnerability in the AI Engine WordPress plugin that allows unauthenticated attackers to delete other users' uploaded chatbot files. The vulnerability affects all versions of the AI Engine plugin before 3.6.4, developed by Jordy Meow. It was publicly disclosed on August 3, 2026, and assigned a CVSS score of 4.8 (Medium) (WPScan, GitHub Advisory).

The root cause is an improper authorization check (CWE-639: Authorization Bypass Through User-Controlled Key) when processing chatbot file deletion requests. The plugin authorizes file deletion actions solely based on a client-supplied session cookie value without verifying that the requesting session actually owns the targeted file. An attacker who obtains a victim's session identifier (e.g., through session hijacking or network interception) and knows the file reference can craft a deletion request impersonating the victim's session, causing the plugin to delete the victim's uploaded files without any server-side ownership validation (WPScan). The vulnerability is classified as OWASP Top 10 A5: Broken Access Control.

Successful exploitation allows an unauthenticated attacker to permanently delete chatbot files uploaded by other guest users, resulting in an availability and integrity impact on victim-owned data. The attack is limited to deletion of files associated with guest chatbot sessions and does not appear to enable remote code execution, data exfiltration, or privilege escalation. The scope is constrained to sites running the AI Engine plugin with guest chatbot file upload functionality enabled (WPScan, GitHub Advisory).

There is no evidence of active in-the-wild exploitation or a publicly available proof-of-concept at this time; WPScan has indicated the PoC will be published on August 17, 2026, to allow time for users to update (WPScan). The EPSS score is approximately 0.00136, reflecting a low probability of exploitation in the near term. The vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog. Exploitation requires the attacker to first obtain a victim's session identifier and a valid file reference, which raises the bar for opportunistic attacks.

1. Session Acquisition: Obtain a victim guest user's session cookie value through network interception (e.g., on an unencrypted connection), session fixation, or other means.
2. File Reference Discovery: Identify the file reference (e.g., filename or file ID) associated with the victim's uploaded chatbot file, potentially through enumeration or prior knowledge.
3. Craft Deletion Request: Send an HTTP request to the AI Engine plugin's file deletion endpoint, supplying the victim's session cookie and the target file reference in the request parameters.
4. Achieve Unauthorized Deletion: Because the plugin validates the action solely against the client-supplied session cookie without verifying file ownership, the server processes the deletion and removes the victim's uploaded chatbot file (WPScan).

• Logs: WordPress access logs showing file deletion requests to the AI Engine chatbot endpoint originating from IP addresses inconsistent with the session owner's known IP; repeated deletion requests across multiple session identifiers from a single source IP.
• File System: Unexpected disappearance of guest-uploaded chatbot files from the AI Engine upload directory without corresponding user-initiated deletion actions.
• Network: HTTP requests to the AI Engine file deletion endpoint carrying session cookies that do not match the originating client's prior session activity.

The vendor has released a patched version: AI Engine 3.6.4 or later, which implements proper server-side ownership verification before processing file deletion requests (WPScan, GitHub Advisory). Site administrators should update the plugin immediately via the WordPress dashboard. If immediate updating is not possible, consider disabling the guest chatbot file upload feature or the plugin entirely as a temporary workaround. Additionally, enforcing HTTPS across the site reduces the risk of session cookie interception that is a prerequisite for exploitation.

The vulnerability was discovered and reported by researcher Revanth Hari Narayana Matte and verified by WPScan (WPScan). No significant broader media coverage or notable community commentary has been identified at this time.