New Delhi: OpenAI has introduced Lockdown Mode, an advanced security setting for ChatGPT designed to reduce the risk of data theft from prompt injection attacks and other emerging AI-related cyber threats. The feature limits ChatGPT’s access to external networks and web-based services, offering additional protection for users and organizations handling sensitive information.
New Feature Targets Prompt Injection Risks
Cybersecurity concerns around artificial intelligence have grown as users increasingly rely on ChatGPT and similar tools for personal, professional and business tasks. Security researchers have warned that prompt injection attacks can manipulate AI systems into revealing or mishandling protected information.
FCRF’s Flagship Cyber Law Certification Returns With a New Four-Week Cohort
Such attacks may involve hidden instructions embedded in websites, emails, documents, images or other digital content. When an AI system processes that content, it may unknowingly follow the attacker’s directions, potentially exposing sensitive data or performing unintended actions.
Lockdown Mode is designed to reduce this risk by tightly limiting how ChatGPT interacts with external systems. OpenAI has said the feature is intended for security-conscious users and organizations that require stronger protection against advanced threats.
External Access Restricted Under Lockdown Mode
Once enabled, Lockdown Mode restricts ChatGPT’s network-enabled capabilities. Live web browsing, Deep Research, Agent Mode, external file downloads and some web-derived image support may be limited or unavailable while the setting is active.
Internet access within Canvas environments is also restricted, preventing code from retrieving information from external online sources. Certain connectors, shopping agents, finance-related agents and other external-service interactions may also be disabled.
Cybersecurity professionals say these restrictions are deliberate. By reducing the number of ways an AI system can communicate with external services, the feature narrows the attack surface available to malicious actors. OpenAI has also clarified that Lockdown Mode does not eliminate every prompt injection risk, but adds a stronger defensive layer against data exfiltration.
Available Across ChatGPT Account Types
Despite the restrictions, several core ChatGPT functions continue to operate. Users can still upload files, use memory features, share conversations and continue ordinary interactions that do not require restricted external access.
OpenAI has made Lockdown Mode available to logged-in users across account types and workspaces. Personal users can enable it through the Security section in Settings, while workspace administrators can configure access for members through workspace controls.
The launch reflects a broader shift in the AI industry, where data protection is becoming central to product design. As AI tools become more widely used in corporate, government and research environments, security features such as Lockdown Mode are expected to play a growing role in protecting sensitive information from evolving cyber threats.
