Guardrails are essential to ensure AI serves the purpose of innovative care delivery models without unintended consequences.