TL;DR: OpenAI has rolled out Lockdown Mode to ChatGPT accounts to mitigate data exfiltration risks from prompt injection attacks.
Summary: The new security feature limits outbound network requests that could be used by an attacker to exfiltrate sensitive data during a prompt injection. While it does not prevent prompt injections from occurring in processed content (like uploaded files or cached web pages), it disrupts the final stage of data theft. The feature is rolling out to Free, Plus, Pro, and self-serve ChatGPT Business accounts.
Why it matters: It establishes a critical network-level defense boundary for applications built on or interacting with ChatGPT data. Developers should monitor how these outbound restrictions affect custom GPTs and integrations that rely on external API calls.
Source: simonwillison.net