Breaking

Therapists build AI ‘twins’ of their own practice as liability and access debates intensify

September 13, 2026 at 13:55 EDT

Healthcare
AI Agents
Policy & Regulation

Some licensed clinicians are cloning their own style into chatbots so patients can get 24/7 replies, a shift that could ease shortages even as it collides with safety, impersonation, and who is legally responsible when an AI agent gives harmful advice.

September 2026 · Mental health AI

Therapist AI twins, always on

Clinicians are cloning their own voice into chatbots so patients can get replies after hours. Access rises. Crisis handling and legal responsibility do not automatically follow.

Crisis cues: humans vs chatbots

Share of appropriate responses to warning signs in a 2025 multi-university comparison. Column height is proportional.

93%

Human therapists

<60%

AI chatbots

Takeaway: the gap is largest where a wrong reply can be dangerous.

51%

Depression-symptom drop after eight weeks with Therabot (no-treatment control)

800M

Reported ChatGPT-class users; 18B messages a week

24/7

Clinician twins reply when the human is off the clock

How a twin is meant to sit in care

1. Capture style
Therapist voice, methods, boundaries

→

2. After-hours chat
Patients message the replica

→

3. Human still owns risk
Crisis, diagnosis, liability unclear

What supporters emphasize

Fills waitlists and nights. Keeps a consistent clinical voice. Some trials show symptom drops versus no care.

What critics flag

Sycophancy, missed suicide cues, competing advice between sessions, impersonation laws, and who pays if the agent errs.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.