Image Credentials: Image Title: OpenAI Introduces Lockdown Mode to Strengthen ChatGPT Security Against Prompt Injection Attacks. Source: (chatgpt.com) Date: June 2026. Attribution: This image was created using AI-generated imagery (chatgpt.com) by Open Chronicle and does not depict a real-world scene.
By Open Chronicle with agencies
OpenAI has unveiled a new security feature called Lockdown Mode for ChatGPT, aimed at protecting users and organizations from one of the most persistent threats facing artificial intelligence systems: prompt injection attacks.
The feature, which is being rolled out across ChatGPT Free, Go, Plus, Pro, and self-serve Business accounts, significantly restricts ChatGPT’s access to external services and the web in order to reduce the risk of sensitive information being exposed through malicious prompts.
According to OpenAI, Lockdown Mode is primarily designed for individuals and organizations that handle confidential data and require stronger safeguards against data exfiltration.
“Lockdown Mode is not intended for everyone. It is designed for people and organisations that handle sensitive data and want stricter protection from data exfiltration risks related to prompt injection,” the company said in its official documentation.
Addressing a Growing AI Security Threat
Prompt injection attacks have become a major concern as AI systems gain broader access to external data sources and online services. These attacks occur when hidden instructions embedded within websites, emails, documents, or images manipulate an AI system into revealing confidential information or performing unintended actions.
OpenAI explained that Lockdown Mode focuses on preventing the final stage of such attacks by limiting outbound network requests that could potentially transfer sensitive data to an attacker.
While the feature reduces the risk of information leakage, the company acknowledged that it does not eliminate prompt injection attempts. Malicious instructions may still appear in content processed by ChatGPT and could influence the quality or accuracy of responses.
Major Restrictions Under Lockdown Mode
Once enabled, Lockdown Mode introduces a series of limitations designed to minimize external exposure.
Live web browsing is restricted to cached content only, reducing the chatbot’s ability to retrieve fresh information from the internet. Deep Research and Agent Mode become entirely unavailable, while code generated within Canvas loses internet access capabilities.
The feature also disables image support, meaning ChatGPT will no longer display images in responses or access image-generation tools while Lockdown Mode remains active.
Several integrated experiences are also affected. Financial tools, shopping-related features, live connectors, and write actions are blocked or unavailable. ChatGPT is prevented from downloading external files for analysis, and networking capabilities inside Canvas-generated code are completely disabled.
Developer Mode cannot operate simultaneously with Lockdown Mode. Activating the security feature automatically disables developer functionality.
Features That Remain Unchanged
Despite the restrictions, several core ChatGPT capabilities remain available.
OpenAI confirmed that Lockdown Mode does not affect memory functionality, file uploads, conversation sharing, or the company’s data usage policies regarding model improvement.
This allows users to maintain their normal workflow while benefiting from enhanced security protections.
How to Enable Lockdown Mode
Users who have access to the feature can activate it through ChatGPT’s settings menu.
To enable Lockdown Mode:
- Open the ChatGPT application or website.
- Navigate to Settings.
- Select Security.
- Open Advanced Security.
- Toggle Lockdown Mode on.
The feature is expected to become increasingly important as AI systems gain deeper integration with external services and personal data, highlighting the growing balance between convenience, functionality, and cybersecurity in the age of artificial intelligence.
A Sign of AI’s Security Evolution
The launch of Lockdown Mode reflects the broader challenges facing the AI industry as advanced language models become more capable and connected. While AI assistants offer unprecedented productivity benefits, they also create new attack surfaces for cybercriminals and malicious actors.
By restricting external access and limiting potential pathways for data leakage, OpenAI’s latest security initiative represents a significant step toward making AI systems safer for enterprise environments, government institutions, researchers, and users handling sensitive information.
As prompt injection attacks continue to evolve, security-focused features such as Lockdown Mode are likely to become an increasingly important part of the AI ecosystem.