OpenAI announces new Lockdown Mode to prevent prompt injection attacks

OpenAI has announced a new feature aimed at providing additional protection against prompt injection attacks, where malicious instructions for chatbots are embedded in web pages and other content sources.

The new Lockdown Mode will disable live web browsing, meaning users can only access cached content. It will also prevent the retrieval and display of images from the web (although users can still generate images) and limit deep research and agent mode functionality.

However, OpenAI cautions that even with Lockdown Mode activated, ChatGPT may still be vulnerable to prompt injections originating from cached web content or uploaded files, which could affect the behavior or accuracy of its responses.

The primary objective of Lockdown Mode is to reduce the risk of sensitive data being inadvertently shared.

OpenAI specifies that “Lockdown Mode is not intended for everyone.” It is designed specifically for individuals and organizations that handle sensitive data and seek enhanced protection against data exfiltration risks associated with prompt injections.

READ MORE: Trump administration explores taking an equity stake in OpenAI

The company is currently rolling out Lockdown Mode to self-serve ChatGPT Business accounts, as well as to eligible personal accounts.