This Sunday morning, OpenAI has launched "Lockdown Mode" for ChatGPT, a critical security update confirmed by The News International, Yellow.com, and entARABI. This new, optional setting severely limits ChatGPT's connections to external services, specifically targeting prompt injection attacks where malicious instructions try to make the AI leak sensitive data. Both The News International and entARABI detail that live web browsing is replaced with cached content, deep research and agent modes ar