ChatGPT Lockdown Mode: New Security for Sensitive Data

OpenAI has launched a new security feature called Lockdown Mode for ChatGPT. This mode aims to reduce the risk of sensitive information being exposed through prompt injection attacks. It's rolling out to eligible personal accounts, including Free, Go, Plus, and Pro users, as well as self-serve ChatGPT Business accounts. Lockdown Mode restricts capabilities that connect ChatGPT to the web or external services. OpenAI states it's for users handling sensitive information who need stronger safeguards against data exfiltration. The company explains that prompt injection attacks involve hidden instructions that try to manipulate AI system behavior. While Lockdown Mode doesn't completely eliminate prompt injection, it prevents the final stage of an attack by limiting outbound network requests that could transfer sensitive information. When enabled, features like live web browsing, deep research, and file downloads for data analysis are disabled. Web searches are limited to cached content, which might be incomplete. This matters because it offers a new layer of protection for those concerned about data privacy when using AI tools.