ChatGPT deploys lockdown mode across plans to block prompt injection
ChatGPT just dropped a new feature called Lockdown mode, now available on all plans, including Free and Business.
Originally launched for ChatGPT for Enterprise, Edu, Healthcare, and Teachers users in February, this mode is designed to combat data theft through prompt injection attacks by limiting the AI's access to live data.
Basically, it's a way of keeping your info safer when you're chatting with the bot.
Lockdown mode uses cached content only
Turning on Lockdown mode means ChatGPT can't browse the live web or pull in real-time images. It sticks to cached content only.
Features like deep research, Agent mode, Canvas networking, and file downloads are also off-limits while it's active.
If you handle sensitive stuff or just want more privacy, you can enable it in your Security settings (just know you'll trade some functionality for that extra layer of protection).