CHATGPT'S LOCKDOWN MODE

Summary from the TLDR AI newsletter

OpenAI introduced an optional Lockdown Mode for higher-risk workflows and added “Elevated Risk” labels for capabilities in ChatGPT, ChatGPT Atlas, and Codex that can increase exposure to prompt-injection attacks.