FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Counsellors and psychotherapists
When a client uses an AI chatbot for emotional support, explore it openly rather than banning or endorsing it, and keep your own risk assessment current. Chatbot advice can conflict with the therapy, and UKCP warns that people harmed by AI may have no professional complaints route.
Part of: AI in counselling and psychotherapy
In short
Some clients may talk to AI chatbots between sessions, sometimes daily. That can help or harm. It may offer comfort at night, but it can also give advice that conflicts with the therapy, reinforce unhelpful beliefs or miss signs of risk. The therapist is not responsible for a chatbot they did not recommend, but they remain responsible for their own work: noticing the effect, discussing it, and reassessing risk. UKCP points out that complaints processes are built around a person's breach, so a client harmed by a chatbot may have no professional body to turn to.
Clients may mention that they have "talked it through" with an AI chatbot before a session, or that a chatbot suggested a diagnosis, a confrontation with a family member, or ending therapy. Some use apps designed for wellbeing; others use general-purpose chatbots that were not built for mental health at all.
UKCP's feature on AI and psychotherapy makes a point that is easy to miss: professional complaints processes are built around a practitioner's breach. If a chatbot gives harmful advice, there may be no professional redress route for the person harmed. That gap shapes how a therapist should respond when a client relies on one.
Treat chatbot use like any other significant relationship or source of advice in the client's life. Useful questions include what they use it for, when, how it makes them feel, and whether anything it said has stayed with them. Some clients feel embarrassed or expect disapproval, so a neutral, curious stance helps.
Look out for heavy reliance, withdrawal from human contact, advice that cuts across agreed safety plans, and chatbot responses that seem to validate self-harm or harmful beliefs. Where risk is present, your own assessment and safety planning come first, whatever the chatbot has said.
These duties apply whether or not AI is involved:
The 2.1(e) risk-assessment duty applies to tools you use or recommend. A client's independent choice of chatbot is not your tool, but if you encourage its use, the position changes: see recommending AI mental health apps.
Illustrative scenario: a client mentions in passing that a chatbot has been "helping me plan" how to tell their partner they are leaving. The counsellor does not explore it. Over the following weeks the client's risk rises, partly because the chatbot reinforced a plan the client was not ready for. After a crisis, the family complains that the counsellor knew about the chatbot and did nothing. The notes record the mention but no discussion or risk review.
The likely complaint is not that the therapist failed to control the chatbot, but that they failed to notice and respond to a known risk factor.
A claim or complaint alleging a failure to assess or manage risk is the core of what therapist PI usually covers, subject to the policy terms. Some PI wordings exclude bodily injury, while therapist malpractice cover may be designed to include injury arising from your professional services; see bodily injury exclusions in PI and check your wording. Prompt notification of any incident that might lead to a claim matters.
Useful controls include:
Apex arranges PI for therapists. A broker can explain how your policy deals with injury-related claims and what to notify. Related guides: AI and suicide risk assessment, AI mental health chatbots in occupational health and the counselling AI hub.
A blanket ban is unlikely to work and may stop clients talking about it. A better approach is to explore what they use, why and how it affects them, and to address any risk or conflicting advice directly. Your own risk assessment and safety planning should come first.
Generally you are not responsible for a tool you did not choose or recommend. You are responsible for your own work, including noticing and responding to risks the client tells you about. If you recommend a tool, you take on more responsibility for that choice.
Yes, where it is relevant to their wellbeing or risk. BACP 3.2(a) requires records that are factual, adequate and relevant. Note what the client said, how you explored it and any change to your risk assessment or safety planning.
UKCP has pointed out that complaints processes are built around a person's breach, so a client harmed by an AI therapy tool may have no professional redress route. They may have other routes, such as consumer or data protection complaints, depending on the circumstances.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
