On June 6, 2026, OpenAI announced that it would fully open ChatGPT's 'Lock Mode' to all individual users (including Free, Go, Plus, and Pro users) and self-service ChatGPT Business accounts. This feature reduces the risk of data leakage by restricting high-risk internet-connected functions such as real-time web browsing, deep research, and Agent mode, allowing only access to cached content and supporting image upload and generation. OpenAI emphasized that Lock Mode cannot completely defend against prompt injection attacks but provides users with a more secure usage environment.
