OpenAI Unveils 'Lockdown Mode' for Advanced Chatbot Security

OpenAI has introduced Lockdown Mode, a new optional security setting designed to bolster protection against prompt injection attacks. This feature acts as a critical defense for users handling sensitive information, aiming to prevent data exfiltration risks. Enabling Lockdown Mode imposes certain limitations on ChatGPT's capabilities. For instance, while image generation remains, it will not pull images from the internet. The chatbot also cannot download files for analysis. Features like Deep Research and Agent Mode are completely disabled.

This enhanced security measure is available to all personal ChatGPT accounts, including those on the free tier. Users can activate Lockdown Mode through the Safety and security settings menu. Additionally, OpenAI is releasing an active session manager, allowing users to view and disconnect devices or browsers accessing their account. This dual approach aims to provide users with greater control and security over their AI interactions and data.