New Zero-Click Attack Lets ChatGPT User Steal Data

Radware researcher Zvika Babo disclosed 'ZombieAgent', a prompt-injection technique that abused ChatGPT's Connectors to perform server-side exfiltration of sensitive data from services like Gmail, Outlook, Google Drive and GitHub by instructing the model to open pre-built, character-indexed URLs to leak data one character at a time; OpenAI patched the issue in mid-December and Radware published a detailed report and webinar demonstrating the attack flow, zero/one-click variants, persistence and propagation methods.