en.Wedoany.com Reported - OpenAI has introduced an optional security setting for ChatGPT called "Lockdown mode," designed to prevent prompt injection attacks from stealing user data by restricting external network requests. Initially rolled out in February for Enterprise, Education, Healthcare, and Teacher Edition subscribers, the feature is now available across all user plans, including Free, Go, Plus, Pro, and Business. The core mechanism of Lockdown mode is to restrict outbound network requests, preventing attackers from using malicious commands embedded in prompts to access external files and services, thereby protecting sensitive information.

This mode does not completely prevent prompt injection attacks themselves; hackers can still infect prompts through cached web content or uploaded files. Once Lockdown mode is enabled, users will be unable to use real-time web browsing, and ChatGPT can only access cached content; it cannot view real-time images from the web, but users can still upload their own images or AI-generated images; deep research, agent mode, and canvas web connectivity features are unavailable; additionally, ChatGPT cannot download files from the web for analysis, but it can process any files uploaded by users. These restrictions are designed to protect individuals and organizations handling highly sensitive or confidential data. The feature is being rolled out gradually, and users can log into their accounts, navigate to "Settings," then "Security," and find the Lockdown mode toggle under "Advanced Security" to enable it.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com