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