OpenAI rolls out Lockdown Mode to protect against prompt injection attacks

3 min readJun 6, 2026 10:58 AM IST

When enabled, Lockdown Mode restricts several ChatGPT capabilities. Users can still upload images and generate images within ChatGPT, but the system may not retrieve images from the internet or display images within responses. (Image: Reuters)

OpenAI has introduced a new security feature called Lockdown Mode, offering users an additional layer of protection against prompt injection attacks, a growing threat in the age of AI-powered assistants.

The company describes Lockdown Mode as an optional security setting primarily for individuals and organisations that handle sensitive information and want stronger safeguards against attempts to exfiltrate data via malicious prompts.

Prompt injection attacks are a form of social engineering in which attackers hide instructions inside webpages, documents or other content processed by AI systems. These instructions are designed to manipulate an AI model into revealing information or performing actions that were not intended by the user.

According to OpenAI, Lockdown Mode acts as a final layer of defence on top of the protections already built into ChatGPT, its AI models and backend infrastructure.

The company notes that the feature is not intended for most users. Instead, it is designed for situations where security takes priority over convenience and access to advanced features.

When enabled, Lockdown Mode restricts several ChatGPT capabilities. Users can still upload images and generate images within ChatGPT, but the system may not retrieve images from the internet or display images within responses. The chatbot is also prevented from downloading files for analysis, although users can continue uploading files directly for review.

Some advanced features are disabled entirely while Lockdown Mode is active. These include Deep Research and Agent Mode, both of which involve broader access to online information and external actions.

Story continues below this ad

OpenAI clarified that the feature does not prevent prompt injections from appearing in content processed by ChatGPT. Instead, it reduces the likelihood that an attacker can exploit network requests or connected tools to extract sensitive information from a user’s account.

The company also said Lockdown Mode does not affect conversation memory, file uploads, conversation sharing settings or whether conversations may be used to improve AI models. In enterprise environments, many of these controls remain configurable by workspace administrators.

Alongside Lockdown Mode, OpenAI is introducing an active session manager that allows users to view devices and browsers currently signed into their account. Users can remotely log out of individual sessions or all sessions if they suspect unauthorised access.

Lockdown Mode is available to all ChatGPT users, including those on the free tier. It can be enabled through the Safety and Security section of ChatGPT settings.