OpenAI Help: Lockdown Mode

This article discusses OpenAI's new Lockdown Mode, a security feature rolling out to eligible ChatGPT accounts (Free, Go, Plus, Pro, and self-serve Business). Lockdown Mode aims to prevent the final stage of data exfiltration during prompt injection attacks by limiting outbound network requests that could transfer sensitive data to attackers. It does not prevent prompt injections from appearing in content but blocks exfiltration vectors. The article explains how this addresses the 'Lethal Trifecta' of LLM security risks and notes tradeoffs in functionality for users with elevated risk profiles. It also includes a tweet from OpenAI's CISO clarifying the feature's intended use.