ShiZhen AI

Jun 8, 2026 · Information Security

Enable ChatGPT’s Lockdown Mode to Prevent Sensitive Data Leaks

OpenAI’s new Lockdown Mode disables network access and external actions in ChatGPT to block prompt‑injection attacks that could exfiltrate private information, trading off features like real‑time browsing, Deep Research, and Agent tasks, and is best used only for handling confidential documents.