CVE-2025-63665: GT Edge AI RCE Vulnerability

CVE-2025-63665 Overview

CVE-2025-63665 is a critical code injection vulnerability affecting GT Edge AI Community Edition versions prior to v2.0.12. The flaw allows remote attackers to execute arbitrary code by injecting a specially crafted JSON payload into the application's Prompt window. This vulnerability represents a significant security risk for organizations deploying GT Edge AI in their AI/ML infrastructure.

Critical Impact

Remote attackers can achieve arbitrary code execution without authentication by exploiting the Prompt window's improper handling of JSON input, potentially leading to complete system compromise.

Affected Products

• GT Edge AI Community Edition versions before v2.0.12
• gtedge gt_edge_ai (CPE: cpe:2.3:a:gtedge:gt_edge_ai:*:*:*:*:community:*:*:*)

Discovery Timeline

• 2025-12-19 - CVE-2025-63665 published to NVD
• 2026-01-05 - Last updated in NVD database

Technical Details for CVE-2025-63665

Vulnerability Analysis

This vulnerability is classified as CWE-94 (Improper Control of Generation of Code - Code Injection). The GT Edge AI Community Edition fails to properly sanitize and validate user-supplied input within the Prompt window interface. When a user submits a JSON payload through this interface, the application processes the data without adequate input validation, allowing attackers to inject and execute arbitrary code on the underlying system.

The attack can be executed remotely over the network without requiring authentication or user interaction, making it highly exploitable. Successful exploitation grants attackers complete control over the affected system, including the ability to read sensitive data, modify configurations, and execute malicious commands with the privileges of the application process.

Root Cause

The root cause of CVE-2025-63665 lies in insufficient input validation and sanitization within the Prompt window's JSON processing functionality. The application directly interprets and executes portions of user-supplied JSON data without properly escaping or validating the content, creating a direct code injection pathway. This represents a fundamental failure in input handling that allows untrusted data to be treated as executable code.

Attack Vector

The attack vector for this vulnerability is network-based, requiring no authentication or user interaction. An attacker can craft a malicious JSON payload containing embedded code and submit it through the Prompt window interface. The vulnerability exploitation flow typically involves:

1. Identifying an accessible GT Edge AI Community Edition instance
2. Crafting a JSON payload with embedded malicious code
3. Submitting the payload through the Prompt window
4. The application processes the JSON without proper sanitization
5. The injected code executes with application privileges

Technical details and proof-of-concept information are available in the GitHub vulnerability research documentation and GitHub Gist PoC.

Detection Methods for CVE-2025-63665

Indicators of Compromise

• Unusual JSON payloads containing code constructs or escape sequences in Prompt window logs
• Unexpected process spawning or command execution originating from the GT Edge AI application
• Anomalous network connections initiated by the GT Edge AI service
• Modifications to application configuration files or system files by the application process

Detection Strategies

• Monitor application logs for malformed or suspicious JSON input patterns submitted to the Prompt window
• Implement network intrusion detection rules to identify potential code injection payloads targeting GT Edge AI
• Deploy endpoint detection and response (EDR) solutions to detect anomalous behavior from the GT Edge AI process
• Review web application firewall (WAF) logs for requests containing potential injection patterns

Monitoring Recommendations

• Enable verbose logging for the GT Edge AI Prompt window interface to capture all submitted input
• Configure alerts for any child process creation by the GT Edge AI application
• Monitor file system changes in the application directory and sensitive system locations
• Implement real-time monitoring of network connections initiated by the application

How to Mitigate CVE-2025-63665

Immediate Actions Required

• Upgrade GT Edge AI Community Edition to version v2.0.12 or later immediately
• If immediate patching is not possible, restrict network access to the GT Edge AI interface to trusted networks only
• Implement input validation at the network perimeter using WAF rules
• Review system logs for any indicators of previous exploitation attempts

Patch Information

The vulnerability is resolved in GT Edge AI Community Edition version v2.0.12 and later. Organizations should upgrade to the latest available version to ensure protection against this vulnerability. Refer to the official GT Edge AI release notes for upgrade instructions and additional security improvements.

Workarounds

• Restrict access to the GT Edge AI Prompt window interface using network segmentation or firewall rules
• Deploy a web application firewall (WAF) with rules to block suspicious JSON payloads
• Implement authentication and access controls in front of the GT Edge AI interface if not already present
• Consider disabling the Prompt window functionality if not required for operations until patching is complete