This isn't specifically a Grok problem, this is an LLM training problem. Internet archive captures some personal info -> LLMs train on it -> you get your data scrubbed from IA, but forget to remove it from LLMs (will Google or OpenAI even respond to your email without legal letterhead..?)