OpenAI Introduces ChatGPT Lockdown Mode to Reduce Data Exfiltration Risks

OpenAI has begun rolling out a new Lockdown Mode for eligible ChatGPT users, designed to reduce the risk of sensitive data exposure through prompt injection attacks. The optional security setting restricts features that can interact with external services, including live web browsing, deep research, agent mode, image retrieval, file downloads, and network-enabled code execution. Rather than preventing prompt injections outright, the feature aims to limit potential pathways for data exfiltration by reducing outbound network access and other high-risk capabilities. OpenAI has also introduced enhanced session management tools, allowing users to review active logins and remotely sign out of devices if unauthorized account activity is suspected.

Reference: thehackernews.com