Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2025-13380 Overview
CVE-2025-13380 is an arbitrary file read vulnerability in the AI Engine for WordPress plugin (also known as liquid-chatgpt). The flaw affects all versions up to and including 1.0.1. The plugin fails to validate user-supplied file paths in the lqdai_update_post AJAX endpoint and uses file_get_contents() with user-controlled URLs in the insert_image() function without restricting protocols. Authenticated attackers with Contributor-level access or higher can read arbitrary files from the server, including sensitive configuration data such as wp-config.php.
Critical Impact
Authenticated attackers with Contributor privileges can read arbitrary files on the WordPress host, exposing database credentials, authentication keys, and other sensitive server-side data.
Affected Products
- AI Engine for WordPress: ChatGPT, GPT Content Generator plugin (liquid-chatgpt)
- All versions up to and including 1.0.1
- WordPress installations with the plugin active and Contributor-level accounts
Discovery Timeline
- 2025-11-25 - CVE-2025-13380 published to NVD
- 2026-06-17 - Last updated in NVD database
Technical Details for CVE-2025-13380
Vulnerability Analysis
The vulnerability is categorized under [CWE-73]: External Control of File Name or Path. The plugin exposes an AJAX endpoint, lqdai_update_post, that accepts user-supplied file path parameters without validation. A separate function, insert_image(), calls PHP's file_get_contents() on URLs provided by the attacker without restricting the URL scheme. Because PHP's file_get_contents() supports stream wrappers such as file://, an attacker can supply a local path and coerce the server into reading arbitrary files accessible to the web server process.
Successful exploitation returns file content back through the plugin workflow, allowing attackers to retrieve sensitive data including wp-config.php, SSH keys, /etc/passwd, and application logs. The authentication requirement limits the attack surface to users with Contributor privileges or higher, which is a low bar on sites that permit guest authoring or community contributions.
Root Cause
The root cause is insufficient input validation combined with unrestricted PHP stream wrapper usage. The lqdai_update_post handler does not sanitize or constrain path input, and insert_image() does not enforce an http:// or https:// protocol allow-list before invoking file_get_contents(). See the plugin source at liquid-chatgpt.php line 315, line 423, and line 83.
Attack Vector
The attack is remote and network-based. An authenticated Contributor sends a crafted AJAX request to the lqdai_update_post endpoint containing a file:// URL or absolute server path. The plugin fetches the referenced content and returns it in the response. Technical analysis is available in the Ryan Kozak CVE write-up and the public PoC repository.
For exploitation specifics, refer to the referenced PoC and advisory. No synthetic code is reproduced here.
Detection Methods for CVE-2025-13380
Indicators of Compromise
- POST requests to admin-ajax.php with the action=lqdai_update_post parameter originating from Contributor-level accounts.
- Request bodies containing file://, absolute filesystem paths (for example /etc/passwd, wp-config.php), or non-HTTP URL schemes.
- Unexpected outbound reads or responses containing filesystem content such as PHP configuration constants or Unix passwd format.
- Newly created Contributor accounts followed by immediate access to the plugin's AJAX endpoints.
Detection Strategies
- Review WordPress access logs for AJAX requests invoking the lqdai_update_post or insert_image actions with suspicious URL parameters.
- Deploy a web application firewall (WAF) rule to block requests to the plugin endpoint containing file://, php://, or local path patterns.
- Monitor PHP error and access logs for file_get_contents() stream wrapper usage on unexpected URIs.
Monitoring Recommendations
- Enable WordPress audit logging to track Contributor and Author account activity, especially plugin AJAX interactions.
- Alert on file reads of sensitive artifacts such as wp-config.php, .env, id_rsa, and /etc/passwd by the web server user.
- Correlate new low-privilege account registrations with subsequent AJAX endpoint activity to identify opportunistic exploitation.
How to Mitigate CVE-2025-13380
Immediate Actions Required
- Deactivate and remove the liquid-chatgpt plugin until a patched version is published and verified.
- Audit WordPress user accounts and remove unused Contributor, Author, or Editor accounts that could be leveraged for exploitation.
- Rotate secrets stored in wp-config.php, including database credentials and WordPress authentication keys, if exploitation is suspected.
- Review web server and WordPress logs for signs of prior exploitation targeting the vulnerable AJAX endpoints.
Patch Information
At the time of publication, the vendor advisory tracked by Wordfence indicates the vulnerability is present in all versions up to and including 1.0.1. Administrators should monitor the plugin's WordPress.org page for a fixed release and apply it immediately upon availability.
Workarounds
- Restrict user registration and limit Contributor-level and higher accounts to trusted personnel only.
- Add WAF rules to block admin-ajax.php requests where the action parameter equals lqdai_update_post and the payload contains non-HTTP URL schemes or local filesystem paths.
- Configure PHP to disable risky stream wrappers by setting allow_url_fopen=Off where application functionality permits.
- Apply filesystem permissions ensuring the web server user cannot read sensitive files outside the WordPress document root.
# Example php.ini hardening to restrict file_get_contents() stream wrappers
allow_url_fopen = Off
open_basedir = "/var/www/html:/tmp"
disable_functions = "file_get_contents_url"
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

