When we published the first edition of AI Adoption in Mental Health Services in 2025, many organisations were still debating whether artificial intelligence was something to embrace or avoid.

One year later, that question has largely been answered. AI adoption is already happening.

Clinicians and other staff are using AI to support their work. Administrators are using it to improve productivity. People seeking care are increasingly turning to AI for information, guidance and support.

But in many organisations, adoption has been driven by individuals rather than the organisation itself. This Shadow AI means leaders may not know which tools are being used, what information is being shared with them, or whether they are being used safely and effectively.

The result is what we call the Confidence Gap: a growing gap between the extent to which AI is being used and organisations' confidence in how it is being used.

Our 2026 edition draws on another year working with mental health organisations implementing AI, as well as live polling conducted during our keynote at TheMHS Conference 2026. While AI use among respondents was widespread, only 8% reported high confidence that it was being used effectively.

The paper explores how organisations can close this gap and move from reactive oversight to deliberate, organisation-led adoption.

We retain the risk and return framework from our first edition and add a new lens: AI autonomy, helping organisations decide both where to use AI and how independently it should operate. We also revisit the AI Fabric of systems, governance and people needed to support adoption at scale.

Drawing on real implementations at Fresh Minds, Caraniche, Raise and VOIX, we show how organisations are already achieving material benefits from AI while managing risk and setting clear boundaries around autonomy.

The objective is not AI adoption itself. It is using AI to improve access to and quality of mental healthcare.

Download the Report