Tech

OpenAI Introduces Lockdown Mode to Strengthen Protection Against AI Prompt Injection Attacks

Blog
/
Tech
/
The new security feature is designed for users and organizations handling sensitive information in ChatGPT.
Tobi Active
June 7, 2026
·
2
mins read

OpenAI has announced a new security feature called Lockdown Mode, aimed at providing stronger protection against prompt injection attacks. These attacks occur when hidden instructions are embedded in webpages, documents, or other content sources, potentially influencing how an AI system responds or handles information.

The new mode is specifically designed for individuals, businesses, and organizations that work with sensitive or confidential data and want an extra layer of protection when using ChatGPT.

When Lockdown Mode is enabled, several advanced features are restricted to reduce potential security risks. This includes disabling live web browsing, meaning ChatGPT can only access previously cached content rather than actively browsing the internet. The feature also blocks the retrieval and display of images from the web, although users can still generate images using AI tools.

In addition, Lockdown Mode turns off capabilities such as Deep Research and Agent Mode, both of which allow ChatGPT to perform more complex tasks and interact with external information sources. By limiting these functions, OpenAI aims to reduce opportunities for malicious instructions to influence AI behavior or extract sensitive information.

OpenAI notes that Lockdown Mode is not a complete solution against prompt injection attacks. The company explained that harmful instructions could still exist within cached web content or uploaded files, meaning there is still a possibility that responses could be affected. However, the feature is designed to significantly lower the chances of sensitive data being exposed through these attacks.

According to OpenAI, Lockdown Mode is not intended for everyday users who simply use ChatGPT for general tasks. Instead, it is targeted at professionals and organizations that prioritize security and need stricter safeguards when working with confidential information.

The company has started rolling out Lockdown Mode to self-service ChatGPT Business customers, while eligible personal account users will also gain access as the feature becomes more widely available.

Stories worth your time, 
every single day

Covering entertainment, tech, business, luxury, politics, and everything in between all in one place.