Return to Threats

A Shared ChatGPT Link Could Run Attacker Code in Your Session and Read Private Chats

thehackernews.com 2026-10-31 data leakage Critical

Executive Summary

A malicious ChatGPT share link could execute attacker code within a user's logged-in session and expose private chats. Organizations using ChatGPT or similar LLM applications should care because shared content can become a path to session compromise and data leakage.

What Happened

A malicious ChatGPT share link could run an attacker's code in the logged-in session of anyone who opened it, allowing them to read their private conversations, researchers at the security firm Parameter showed. The researchers reported the flaw through OpenAI's bug bounty program on October 18, and OpenAI fixed it eight days later and rated it P1, its most serious category.

Affected Products

  • ChatGPT

Vulnerability / Attack Technique

Malicious shared-link exploitation

Technical Mechanism

Opening a malicious ChatGPT share link could run attacker code in the logged-in session of the person who opened it. The reported impact included access to that user's private conversations.

AI Security Category

LLM02:2025 Sensitive Information Disclosure

Relevant Frameworks

LLM02:2025 Sensitive Information DisclosureMITRE ATLAS: Initial AccessMITRE ATLAS: ExecutionMITRE ATLAS: CollectionMITRE ATLAS: Exfiltration

RealGround Analysis

This incident shows how an LLM application's sharing feature can create a security boundary failure between user-controlled links, authenticated sessions, and private conversation data. Teams running AI agents or LLM applications should validate that shared content cannot execute code or access data beyond the intended sharing scope.

Recommended Controls

  • Enforce strict isolation between shared-link content and authenticated application sessions.
  • Prevent shared content from executing scripts or attacker-controlled code in the application context.
  • Apply least-privilege access controls to private conversations and session data.
  • Red-team shared-link workflows for cross-user data access and session compromise.
  • Monitor and rapidly remediate vulnerabilities affecting authenticated AI application features.

Related RealGround Service

AI Agent Business Logic Audit · or start with the free AI Agent Security Scanner.

Sources

thehackernews.com: https://thehackernews.com/2026/10/a-shared-chatgpt-link-could-run.html

Summary and analysis written by RealGround from the source above; last updated 2026-10-31.

Talk to AI CISO