A Shared ChatGPT Link Could Run Attacker Code in Your Session and Read Private Chats
Read original article ↗AI Summary
Researchers at Parameter discovered a critical vulnerability in ChatGPT that allowed a maliciously crafted public share link to execute attacker-controlled code within a victim's authenticated session. By exploiting a combination of features—Preview rendering and an insecure message listener from the MathLive library—attackers could bypass sandboxing and inject HTML into the trusted chatgpt.com context, enabling full account takeover. This could allow the reading of private conversations and performing any action on behalf of the victim. OpenAI patched the issue within eight days of disclosure, classifying it as P1, their highest severity level.
AI-extracted · verify before operational use
No entities or IoCs were extracted from this article.