CVE-2025-13374: Arbitrary File Upload enables RCE

CISO Take

CVE-2025-13374 is a critical unauthenticated RCE vulnerability in the Kalrav AI Agent WordPress plugin — any site running version ≤2.3.3 is fully exposed with zero prerequisites for exploitation. An attacker can upload a PHP web shell and achieve full server compromise in under five minutes, with a public PoC already available on GitHub. Immediately disable or remove the plugin; if the server has been internet-exposed since January 2025, treat it as potentially compromised and rotate all AI API keys stored on that host.

What is the risk?

CRITICAL. CVSS 9.8 with network-accessible, zero-authentication, zero-interaction exploitation. The plugin's AJAX handler accepts file uploads without any type or extension validation, meaning any unauthenticated HTTP POST can plant executable code on the server. Exposure is every public-facing WordPress site with the plugin installed — no scanning, fingerprinting, or credential acquisition needed. A working PoC (github.com/d0n601/CVE-2025-13374) is already public, making mass exploitation by unsophisticated attackers highly likely.

How severe is it?

CVSS 3.1

9.8 / 10

EPSS

1.1%

chance of exploitation in 30 days

Higher than 60% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ Public PoC indexed (trickest/cve)

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AV Network

AC Low

PR None

UI None

S Unchanged

C High

I High

A High

What should I do?

7 steps

IMMEDIATE

Enumerate all WordPress installations with kalrav-ai-agent plugin ≤2.3.3 across your environment.
DISABLE or REMOVE the plugin — no patch is available; do not wait.
SCAN webroot and upload directories for recently created .php, .phtml, .phar, or .cgi files (check last 60 days minimum).
AUDIT server logs for POST requests to wp-admin/admin-ajax.php with action=kalrav_upload_file — any hit is a confirmed exploitation attempt.
ROTATE all AI API keys stored on the affected server (OpenAI, Anthropic, database credentials, Stripe keys).
If compromise is confirmed, treat server as fully owned and perform fresh OS-level deployment.
As a temporary WAF control if removal is not immediately possible, block multipart/form-data POST requests to the vulnerable AJAX action endpoint.

What does CISA's SSVC say?

Decision Track*

Exploitation none

Automatable Yes

Technical Impact total

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Code Execution Auth Bypass Data Extraction Plugin Agent API AML.T0010.001 - AI Software AML.T0037 - Data from Local System AML.T0049 - Exploit Public-Facing Application AML.T0050 - Command and Scripting Interpreter AML.T0053 - AI Agent Tool Invocation AML.T0072 - Reverse Shell AML.T0081 - Modify AI Agent Configuration AML.T0083 - Credentials from AI Agent Configuration

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

ISO 42001

A.6.2 - AI system security A.6.2.6 - AI system security and resilience A.8.4 - AI system acquisition, development and maintenance

NIST AI RMF

MANAGE 2.2 - Mechanisms are in place to respond to and recover from incidents MANAGE-2.2 - Mechanisms are in place and applied to sustain the value of deployed AI systems MAP 2.1 - Scientific findings and organizational risks are identified and used

OWASP LLM Top 10

LLM07 - Insecure Plugin Design LLM08 - Excessive Agency

Frequently Asked Questions

What is CVE-2025-13374?

CVE-2025-13374 is a critical unauthenticated RCE vulnerability in the Kalrav AI Agent WordPress plugin — any site running version ≤2.3.3 is fully exposed with zero prerequisites for exploitation. An attacker can upload a PHP web shell and achieve full server compromise in under five minutes, with a public PoC already available on GitHub. Immediately disable or remove the plugin; if the server has been internet-exposed since January 2025, treat it as potentially compromised and rotate all AI API keys stored on that host.

Is CVE-2025-13374 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2025-13374, increasing the risk of exploitation.

How to fix CVE-2025-13374?

1. IMMEDIATE: Enumerate all WordPress installations with kalrav-ai-agent plugin ≤2.3.3 across your environment. 2. DISABLE or REMOVE the plugin — no patch is available; do not wait. 3. SCAN webroot and upload directories for recently created .php, .phtml, .phar, or .cgi files (check last 60 days minimum). 4. AUDIT server logs for POST requests to wp-admin/admin-ajax.php with action=kalrav_upload_file — any hit is a confirmed exploitation attempt. 5. ROTATE all AI API keys stored on the affected server (OpenAI, Anthropic, database credentials, Stripe keys). 6. If compromise is confirmed, treat server as fully owned and perform fresh OS-level deployment. 7. As a temporary WAF control if removal is not immediately possible, block multipart/form-data POST requests to the vulnerable AJAX action endpoint.

What systems are affected by CVE-2025-13374?

This vulnerability affects the following AI/ML architecture patterns: WordPress AI agent deployments, Plugin-based AI chatbot integrations, Web-hosted model inference frontends, AI agent frameworks with web interfaces, Server-side RAG pipelines accessible from compromised WordPress host.

What is the CVSS score for CVE-2025-13374?

CVE-2025-13374 has a CVSS v3.1 base score of 9.8 (CRITICAL). The EPSS exploitation probability is 1.06%.

What is the AI security impact?

Affected AI Architectures

WordPress AI agent deploymentsPlugin-based AI chatbot integrationsWeb-hosted model inference frontendsAI agent frameworks with web interfacesServer-side RAG pipelines accessible from compromised WordPress host

MITRE ATLAS Techniques

AML.T0010.001 AI Software

AML.T0037 Data from Local System

AML.T0049 Exploit Public-Facing Application

AML.T0050 Command and Scripting Interpreter

AML.T0053 AI Agent Tool Invocation

AML.T0072 Reverse Shell

AML.T0081 Modify AI Agent Configuration

AML.T0083 Credentials from AI Agent Configuration

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.6.2, A.6.2.6, A.8.4

NIST AI RMF: MANAGE 2.2, MANAGE-2.2, MAP 2.1

OWASP LLM Top 10: LLM07, LLM08

What are the technical details?

Original Advisory

The Kalrav AI Agent plugin for WordPress is vulnerable to arbitrary file uploads due to missing file type validation in the kalrav_upload_file AJAX action in all versions up to, and including, 2.3.3. This makes it possible for unauthenticated attackers to upload arbitrary files on the affected site's server which may make remote code execution possible.

Exploitation Scenario

An adversary uses WPScan or Shodan to identify WordPress sites running the Kalrav AI Agent plugin. With a single unauthenticated HTTP POST to /wp-admin/admin-ajax.php?action=kalrav_upload_file, they upload a PHP web shell — no extension validation means any filename works. They browse to the uploaded file URL to gain interactive command execution. Within minutes they extract OpenAI/Anthropic API keys from wp-config.php and server environment variables, dump the WordPress database for user PII, establish persistence via a cron-based reverse shell, and pivot to internal AI services reachable from the server. The full chain requires no credentials, no AI/ML knowledge, and under five minutes of effort.

Weaknesses (CWE)

CWE-434 Unrestricted Upload of File with Dangerous Type Primary

CWE-434 — Unrestricted Upload of File with Dangerous Type: The product allows the upload or transfer of dangerous file types that are automatically processed within its environment.

[Architecture and Design] Generate a new, unique filename for an uploaded file instead of using the user-supplied filename, so that no external input is used at all.[REF-422] [REF-423]
[Architecture and Design] When the set of acceptable objects, such as filenames or URLs, is limited or known, create a mapping from a set of fixed input values (such as numeric IDs) to the actual filenames or URLs, and reject all other inputs.

Source: MITRE CWE corpus.