OpenAI introduces lockdown mode in ChatGPT to boost security

OpenAI is rolling out a new 'Lockdown Mode' for ChatGPT, aimed at protecting users and organisations handling highly sensitive information from advanced cyber threats.

The feature is an optional, advanced security setting that tightly restricts how ChatGPT interacts with external systems, reducing the risk of prompt-injection attacks and data leaks. According to the company, Lockdown Mode is designed for high-risk users such as enterprises, researchers, and security-sensitive professionals rather than the general public.

When enabled, the mode disables or limits several capabilities that could expose data. These include restricting web browsing to cached content, turning off deep research and agent-based tools, and reducing access to certain network-connected features. The goal is to prevent attackers from exploiting AI systems to extract confidential information.

Alongside Lockdown Mode, OpenAI is also introducing “Elevated Risk” labels across its products. These in-product warnings highlight features that may carry additional security risks when connected to external apps or the internet, helping users make informed decisions about enabling them.