OpenAI Launches Lockdown Mode to Combat Prompt Injection Attacks

OpenAI has introduced Lockdown Mode, an advanced security feature for ChatGPT, aimed at mitigating prompt injection attacks. This mode limits the AI's ability to connect to the web or external services, significantly reducing the risk of data exfiltration. Prompt injection attacks involve embedding malicious instructions within content that AI systems process, potentially leading to unauthorized data access. Lockdown Mode disables features like live web browsing, Deep Research, and Agent Mode, while still allowing users to upload files. Initially launched for enterprise users in February 2026, it is now available to all personal accounts, including Free, Plus, Pro, and self-serve Business accounts. OpenAI emphasizes that while Lockdown Mode enhances security, it does not completely eliminate the risk of prompt injections. Users are encouraged to enable this feature if they handle sensitive information.

Key Points: • Lockdown Mode limits AI's web connectivity to enhance security against prompt injections. • The feature is now available for all personal ChatGPT accounts and self-serve Business accounts. • Prompt injection attacks can manipulate AI behavior by embedding malicious instructions in processed content.