OpenAI has announced the rollout of a new security-focused feature called Lockdown Mode for ChatGPT, aimed at reducing the risk of sensitive information being exposed through prompt injection attacks. The feature is being made available to eligible personal users across Free, Go, Plus, Pro, and self-service Business plans.
The new setting is designed for individuals and organizations that work with confidential data and require stronger safeguards when using artificial intelligence tools. By restricting access to several web-connected capabilities, Lockdown Mode helps minimize opportunities for attackers to extract information through malicious prompts.
Focused on Limiting Data Exfiltration
Prompt injection attacks remain one of the most challenging security concerns facing large language models. In such attacks, hidden or deceptive instructions attempt to manipulate AI systems into revealing sensitive information or performing unintended actions.
Lockdown Mode does not stop these attacks from occurring. Instead, it aims to block the pathways that attackers could use to move data outside the system. The feature builds on existing security controls and sandboxing measures by limiting outbound network activity that could potentially be exploited for unauthorized data transfers.
According to OpenAI, the security enhancement is intended to significantly reduce the likelihood of prompt injection-based data exfiltration while maintaining core ChatGPT functionality.
Features Restricted Under Lockdown Mode
To strengthen security protections, several ChatGPT capabilities are either disabled or heavily restricted when Lockdown Mode is enabled. These include:
- Live web browsing, which is limited to cached content rather than direct internet access.
- Image support, including image retrieval from the web and image display within responses.
- Deep research tools that rely on broader online access.
- Agent Mode, which can perform more advanced automated tasks.
- Canvas networking, preventing generated code from accessing external networks.
- File downloads, restricting the ability to retrieve files for analysis purposes.
These limitations are designed to reduce the number of potential channels through which sensitive information could leave a protected environment.
Not Intended for Every User
OpenAI emphasized that Lockdown Mode is an optional feature targeted primarily at users with elevated security requirements. Because of its restrictive nature, some advanced functions and convenience features become unavailable while the mode is active.
The company also confirmed that Lockdown Mode cannot be used simultaneously with Developer Mode. Activating one option automatically disables the other to prevent conflicting configurations.
Security Benefits With Remaining Risks
While the new feature improves protection against data leakage attempts, OpenAI acknowledged that no security measure can completely eliminate risk.
The company noted that attackers may still discover new techniques or exploit interactions involving approved applications and unforeseen feature combinations. Additionally, Lockdown Mode does not prevent all consequences of prompt injection attacks.
For example, malicious instructions embedded within uploaded files could still influence how ChatGPT interprets information and generates responses, potentially leading to inaccurate outputs even if data exfiltration is blocked.
New Session Management Tools Also Released
Alongside Lockdown Mode, OpenAI has introduced enhanced account management capabilities that allow users to monitor active ChatGPT sessions more effectively.
The new session management dashboard enables users to review connected devices and sign out remotely from individual sessions or all active sessions if suspicious account activity is detected.
Users can view details such as:
- Device information
- Application used for access
- Approximate location
- Login date and time
- Trusted device status
- Current active session indicators
The addition provides users with greater visibility into account activity and helps strengthen overall account security.
Growing Focus on AI Security
The launch of Lockdown Mode highlights the increasing emphasis on AI security as organizations integrate generative AI into workflows involving sensitive information. As prompt injection techniques continue to evolve, technology providers are investing in layered security measures to reduce potential exposure risks while maintaining the benefits of AI-powered productivity tools.
With cyber threats targeting AI platforms becoming more sophisticated, features like Lockdown Mode represent a proactive step toward improving trust, privacy, and data protection in enterprise and personal AI usage.
