A Shared ChatGPT Link Could Run Attacker Code in Your Session and Read Private Chats — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Parameter researchers showed that opening a malicious ChatGPT share link runs an attacker's code in the logged-in session of whoever clicks it, handing the attacker control inside that session.
- ChatGPT private conversations could be read by the attacker as a consequence, with the exposure tied to the logged-in session of the person who opened the link rather than to the shared conversation alone.
- OpenAI received the flaw report through its bug bounty program on October 18, when Parameter's researchers reported the vulnerability to the company.
Why it matters: For OpenAI, the report landed through its own bug bounty program on October 18, showing researchers had a channel for this class of flaw. For users, the second-order consequence is blunt: attacker code in a logged-in session means the victim's private conversations, not just the shared link, are exposed.
Ask SkimNews
