Artificial intelligence (AI) is increasingly being used in mental health care, with new tools and clinical trials highlighting both opportunities and concerns. The mindfulness app Headspace has launched an AI companion tool called Ebb in the UK, designed to provide sub-clinical support for users by helping them process thoughts and emotions and offering mindfulness recommendations. Ebb does not provide direct mental health guidance or diagnoses, but aims to complement human therapists and offer support when traditional care is unavailable. Headspace's chief clinical officer, Dr. Jenna Glover, stated that Ebb was designed for self-reflection and is available 24/7, with a focus on privacy and data security. Since its US launch in October 2024, Ebb has facilitated over 1.4 million messages, with 64% of users reporting they felt heard and understood. The tool was developed by a cross-functional team, including clinical psychologists and AI specialists, and follows evidence-based principles such as Motivational Interviewing. The first clinical trial of a generative AI therapy bot, Therabot, developed by Dartmouth College’s Geisel School of Medicine, found that people with depression, anxiety, or at risk for eating disorders benefited from interacting with the bot. Researchers emphasized the importance of training AI models on evidence-based data, noting that earlier versions trained on non-clinical data produced inappropriate responses. Therabot's development began in 2019 and required over 100,000 human hours. Cloudflare reported a 251% increase in generative AI service use between February 2024 and March 2025. Many people are turning to AI chatbots such as OpenAI's ChatGPT for therapy, partly due to long NHS mental health wait times. However, experts, including psychotherapist James Lloyd, caution that overreliance on AI for emotional support may reduce essential human social connections. A 2024 Harvard study found that social connection is a strong predictor of long-term physical and mental health, while social isolation increases health risks. Clinicians and technology experts advocate for a balanced approach, using AI to enhance access to mental health support without replacing the vital role of human therapists.