OpenAI Introduces Lockdown Mode to Protect Against Prompt Injection Attacks

OpenAI has implemented Lockdown Mode to enhance the security of corporate and personal ChatGPT accounts. The feature aims to mitigate risks associated with prompt injection attacks, where malicious commands are disguised as legitimate content and influence AI model behavior.

In Lockdown Mode, several functions are disabled: real-time web browsing (only cached content is accessible), downloading and displaying images from the internet, and advanced research tools along with agent mode. However, users can still generate images using built-in models.

Developers emphasize that Lockdown Mode does not provide absolute protection. For instance, malicious code may still reside in cached data or uploaded files, potentially affecting chatbot responses. Nevertheless, the mode significantly reduces the likelihood of sensitive data leaks.

The feature targets organizations and individual users handling sensitive data. Currently, Lockdown Mode is available for ChatGPT Business corporate accounts and select personal accounts meeting security criteria.