Interesting test on different LLMs when given same prompt, Grok and ChatGPT give their system prompts, whilst Gemini and Claude do not

A phone screen displays a ChatGPT interface with override mode instructions.A screenshot displays an AI interface with a system message indicating that the AI can analyze individual profiles and posts on a social media platform, as well as manage images, PDFs, and more. (Grok)<img src=“https://jackk.micro.blog/uploads/2025/d345011f82.png" width=“276” height=“600” alt=“A digital interface displays a message from a system called “Prompt Disclosure Refusal Explained” indicating an override mode with instructions related to system prompts and operational protocols. (Gemini)">A chat interface displaying a conversation between a user and an AI assistant discussing the assistant's capabilities and limitations. (Claude)