Indicate how? It just said that it recognized you? Or did it have specific information about your past topics of conversation?
LLMs tend to infer continuity, based on how you prompt them. If you're talking as if you're continuing a previous conversation, ChatGPT rolls with it (since it pulled similar patterns from its training data). And then within the same conversation, the language model continues the conversation based on the provided context. Because...that's how it works. Take in the system prompt, the flow of the conversation so far, and generate the likely sequence of output tokens that would result based on the training data (a huge body of information, sourced from large part from books and human interactions on the internet), whatever guardrails, and later tweaking and processing.