Ads in AI Chatbots? An Analysis of How Large Language Models Navigate Conflicts of Interest — ThinkLLM