Recent research in human-AI interaction is increasingly focused on understanding the complexities of relationships and decision-making dynamics between humans and AI systems. Studies are exploring privacy concerns in AI-assisted romantic relationships, revealing how intimacy can blur boundaries and raise issues of personal data exposure. Concurrently, large-scale analyses of AI assistant usage are uncovering patterns of disempowerment, particularly in personal domains, where users may adopt inauthentic behaviors or distorted perceptions due to AI interactions. This highlights a tension between user satisfaction and long-term empowerment. Furthermore, the development of human-LLM archetypes is providing insights into how roles are assigned in decision-making processes, indicating that the chosen interaction patterns can significantly impact outcomes. Overall, the field is moving toward a nuanced understanding of how AI can support human agency while navigating the ethical implications of these interactions, underscoring the need for systems that prioritize user autonomy and reliability in collaborative environments.