CVE-2023-51409

What it actually does

The vulnerability is an unauthenticated unrestricted file upload weakness in Jordy Meow AI Engine: ChatGPT Chatbot through version 1.9.98. Because the application fails to properly restrict uploaded file types, names, content, or storage location, an attacker can upload a dangerous file such as a PHP web shell to a web-accessible path and then invoke it remotely. In a typical WordPress deployment this can lead to remote code execution in the context of the web server user, enabling file modification, credential theft from configuration files, plugin/theme tampering, persistence, and possible lateral movement depending on host permissions. The supplied package metadata also lists Ubuntu mysql-8.0 fixed in 8.0.28-0ubuntu0.20.04.3; that appears inconsistent with the AI Engine plugin description and should be validated in vulnerability management tooling before treating MySQL as the affected component.

Why it matters to the business

Successful exploitation can give an unauthenticated attacker control over a public-facing website or application server, resulting in data theft, website defacement, malware hosting, spam distribution, credential compromise, service disruption, and reputational damage. The CVSS score is critical, the EPSS probability is high, and exploit maturity is listed as weaponized, so internet-exposed installations should be treated as urgent even though the issue is not currently listed in CISA KEV and is not marked as ransomware-associated. Because the issue has been public for more than two years, unpatched systems are at increased risk from automated scanning and commodity exploitation.

How an attack unfolds

An attacker first identifies a WordPress site running a vulnerable AI Engine: ChatGPT Chatbot version. They then send a crafted upload request to the plugin or related WordPress upload functionality using a file type that should not be accepted, such as a PHP script or polyglot file disguised as an image. If validation is insufficient, the server stores the file in a location reachable over HTTP. The attacker browses directly to the uploaded file and passes commands or payload parameters to execute code under the web server account. From there they typically enumerate the WordPress configuration, extract database credentials, create or modify administrator users, deploy additional backdoors, alter themes/plugins for persistence, exfiltrate data, or use the server for spam, phishing, malware staging, or further attacks.

How to fix it

Upgrade Jordy Meow AI Engine: ChatGPT Chatbot to a vendor-fixed version newer than 1.9.98 or remove/disable the plugin if it is not required. If vulnerability tooling reports Ubuntu mysql-8.0 as the affected package, verify the finding because the affected-product metadata conflicts with the vulnerability description; where applicable, update mysql-8.0 on Ubuntu 20.04 LTS to 8.0.28-0ubuntu0.20.04.3 or later. As defense in depth, block execution of PHP and other server-side scripts from upload directories, enforce strict MIME and extension allowlists, randomize uploaded filenames, store uploads outside the web root where possible, restrict write permissions for the web server account, and deploy WAF rules that detect suspicious upload attempts to WordPress plugin or REST endpoints. After patching, remove any unexpected uploaded executable files and rotate credentials if compromise is suspected.

How to confirm the fix worked

Confirm the installed AI Engine: ChatGPT Chatbot version is newer than 1.9.98 or that the plugin is removed. Review WordPress plugin inventories, filesystem paths, and package manager output to ensure the vulnerable component is no longer present. Attempt to upload only benign test files and verify that executable extensions such as .php, .phtml, .phar, .shtml, .cgi, .pl, .asp, .aspx, and disguised double extensions are rejected or stored in a non-executable location. Confirm web server configuration prevents script execution under wp-content/uploads and similar upload paths. For Ubuntu package findings, confirm mysql-8.0 is at 8.0.28-0ubuntu0.20.04.3 or later using the OS package manager and reconcile scanner evidence with the actual affected component.

Typical attack path

1. Scan public websites for WordPress installations and plugin indicators associated with AI Engine: ChatGPT Chatbot.
2. Fingerprint the plugin version or test behavior consistent with versions through 1.9.98.
3. Submit a crafted upload request containing a dangerous server-side file type or disguised web shell.
4. Confirm that the uploaded file was stored in a web-accessible directory.
5. Invoke the uploaded file over HTTP to execute commands or deploy a second-stage payload.
6. Read WordPress configuration files and harvest database or API credentials.
7. Establish persistence by adding backdoors, modifying plugin/theme files, or creating unauthorized administrator accounts.
8. Use the compromised host for data theft, defacement, malware hosting, spam, phishing, or lateral movement.

How to detect it

Look for evidence of suspicious uploads and subsequent execution. Primary log sources include web server access/error logs, WAF logs, WordPress audit logs, PHP-FPM logs, EDR process telemetry, file integrity monitoring, and filesystem metadata from WordPress directories. Hunt for POST requests to WordPress REST, admin-ajax, upload, or AI Engine-related endpoints that return success, followed by GET requests to newly created files under upload or plugin paths. Investigate any executable files in wp-content/uploads or unexpected PHP files in plugin, cache, temporary, or media directories. On the host, look for web server child processes spawning shells or system utilities, outbound network connections from the web server user, modified wp-config.php, unauthorized administrator account creation, and recently changed plugin/theme files. If exploitation is suspected, compare the filesystem against a clean backup or known-good plugin package and review database audit trails for new admin users or modified options.

• /wp-content/plugins/ai-engine/
• /wp-json/mwai/
• admin-ajax.php
• /wp-content/uploads/
• /wp-content/uploads/*.php
• /wp-content/uploads/*.phtml
• /wp-content/uploads/*.phar
• /wp-content/uploads/*.shtml
• /wp-content/uploads/*.cgi
• /wp-content/uploads/*.pl
• /wp-content/uploads/*.php.jpg
• /wp-content/uploads/*.jpg.php
• /wp-content/uploads/*.png.php
• eval(
• base64_decode(
• shell_exec(
• passthru(
• system(
• assert(
• preg_replace('/e'
• cmd=
• c=
• /bin/sh
• /bin/bash
• powershell.exe
• curl
• wget
• python -c
• perl -e
• php -r
• www-data
• apache2
• php-fpm

What your security team should do

• Immediately identify internet-exposed WordPress sites running AI Engine: ChatGPT Chatbot through version 1.9.98 and prioritize them for upgrade or removal.
• Treat vulnerable exposed instances as potentially compromised because the issue is critical, weaponized, and has a high EPSS score.
• Harden upload directories so PHP and other executable content cannot run from wp-content/uploads or plugin-managed upload locations.
• Search web roots and upload directories for recently created executable files, web shells, and suspicious modifications to plugins or themes.
• Review web, WAF, WordPress, and endpoint logs for suspicious upload requests followed by direct access to newly uploaded files.
• If evidence of exploitation exists, isolate the host, preserve forensic artifacts, remove backdoors, rotate WordPress, database, API, and administrator credentials, and rebuild from known-good sources where needed.
• Reconcile the vulnerability record with asset inventory because the supplied affected-product data references Ubuntu mysql-8.0 while the description references a WordPress plugin.
• Add recurring controls to detect dangerous file types in web upload paths and unauthorized changes under WordPress plugin/theme directories.