OpenAI announced a new feature that provides additional protection from prompt injection attacks, where malicious chatbot instructions are hidden in web pages or other content sources.
Lockdown mode disables, among other things, live web browsing (you can only access cached content), retrieving and displaying images from the web (although you can generate images), deep exploration, and agent mode.
According to the company, even when lockdown mode is turned on, ChatGPT may still be vulnerable to prompt injections, which, for example, can “appear in cached web content or uploaded files and may affect the behavior or accuracy of responses.”
However, the goal is to reduce the likelihood that sensitive data will be shared in the process.
“Lockdown mode is not for everyone,” OpenAI says. “It is designed for people and organizations that work with sensitive data and require tighter protection from data breach risks associated with instant insertion.”
The company says it is now rolling out lockdown mode to self-service ChatGPT Business accounts and eligible personal accounts.
