The integration of artificial intelligence into mental health practice represents a fundamental shift in how therapeutic care is delivered and documented. As AI tools move from experimental curiosities to clinical realities, understanding how practicing therapists actually perceive and use these technologies becomes crucial for patient safety and treatment effectiveness. This qualitative investigation examined how 18 actively practicing psychotherapists in the United States navigate their relationship with generative AI tools in therapeutic contexts. Through hour-long interviews conducted between January and May 2025, researchers explored the nuanced ways clinicians calibrate trust and skepticism toward AI applications ranging from documentation assistance to client-facing chatbots. The study reveals therapists grappling with competing priorities: efficiency gains from AI documentation tools versus concerns about maintaining the authenticity of therapeutic relationships. Participants demonstrated sophisticated reasoning about when and how to deploy AI assistance, with many expressing particular caution about tools that might interfere with the human connection central to therapeutic outcomes. These findings illuminate a critical tension in modern healthcare delivery. While AI promises to reduce administrative burden and potentially expand access to mental health services, therapists remain deeply protective of the relational foundation that drives therapeutic success. This research provides essential insight into clinician perspectives that will shape responsible AI integration in mental health settings. The study's timing is particularly relevant as healthcare systems worldwide accelerate AI adoption without fully understanding practitioner concerns or implementation challenges in sensitive therapeutic contexts.