OpenAI launches ChatGPT Lockdown Mode to curb data leaks

OpenAI has introduced Lockdown Mode, an optional security setting for ChatGPT that restricts the model’s ability to interact with external resources. When enabled, live web browsing is disabled, cached content only is used, and features such as Deep Research, Agent Mode and network‑enabled code execution are turned off. The mode also blocks file‑download analysis, limiting the AI to working with uploaded images and locally generated visuals. The purpose is to reduce the risk of prompt‑injection attacks that could lead to the exfiltration of sensitive information from organizations handling confidential data. OpenAI notes that Lockdown Mode does not prevent malicious instructions from appearing in content, but it stops the final step of extracting data. The trade‑off is a loss of some of ChatGPT’s most powerful capabilities. The feature mirrors Apple’s own “Lockdown Mode,” designed for users at high risk of sophisticated cyber‑attacks, and signals a broader industry shift toward tighter control over AI assistants’ access to external services.

Lockdown Mode is available to all ChatGPT account types and is aimed at security‑focused professionals and enterprises that need stronger protection for sensitive workloads.