A newly patched security flaw in ChatGPT allowed attackers to run malicious code inside a victim's logged-in browser session. Simply opening a shared chat link let an attacker read private conversations and execute account actions without user interaction.
FireBrief
FireBrief