Last year, It’s Complicated surveyed both sides of the therapy room, clients and practitioners, about how AI was showing up in their care and their work. It’s still rare for research in this space to capture both perspectives together. Most surveys ask one side or the other. Ours asks both, so we can see not just how people are using AI, but how that use is landing on the other side of the relationship too.<\/p>\n\n\n\n
A year on, we want to know what’s changed, and this year there’s a new reason it matters.<\/p>\n\n\n\n
Since 2 August 2026, AI chatbots operating in Europe have been legally required to tell you they’re AI, the same rule that applies to any chatbot, whether it’s tracking a parcel or standing in for a therapist. Self-help mental health chatbots aren’t classified as high-risk under EU law, so that disclosure duty is the only real obligation they’re under. The European Commission’s own review this year found that self-help chatbots have already been linked to real harm in Europe, including documented deaths, and still concluded no further rule change is needed for now. A tool people turn to for something as sensitive as their mental health is currently held to no higher a legal standard than one that helps you find a recipe.<\/p>\n\n\n\n
That gap is part of why this year’s survey matters, but it’s only one piece of what we’re asking. Alongside new questions on regulation and disclosure, this year’s AI in therapy report<\/a> covers how clients and practitioners are actually using AI, what’s helped, what hasn’t, and how that’s shifted since last year’s findings<\/a>.<\/p>\n\n\n\n