AI use policy

This policy sets out how I approach artificial intelligence in my counselling practice — where I use it, where I deliberately do not, and what you can expect from me. It sits alongside my Privacy Notice, my Terms of Use and the client agreement you sign before we begin work together.

It reflects the UK GDPR and core responsibility 2.1(e) of the BACP Ethical Framework for the Counselling Professions 2026.

My core principle

Where AI is, and is not, used in my practice

AI plays a small and deliberately limited part in how I work.

Where I use it

• Drafting worksheets and exercises — for example a thought record, a values exercise, or a sheet on managing panic. I ask for the technique or the topic and nothing else. I do not describe you, your situation, or anything you have told me.
• The assistant on my website, which answers practical questions about fees, location, session formats and booking. It is powered by Anthropic's Claude and is operated on my behalf by Deskcush Ltd, who build and run my website.
• Marketing and social media content — for example captions and posts for Instagram. I write or approve everything before it goes out.
• My self-help workbook, which I sell to the public separately from my counselling practice. See below.
• General content and administration — for example drafting website copy, policies and other practice documents.

The tools I currently use are Anthropic's Claude, Google Gemini and Proton Lumo. None of them receives any information about you. Lumo is Proton's own assistant, and is the most protective of the three: conversations are held under zero-access encryption so that only I can read them, Proton keeps no logs of what is asked or answered, the models run on Proton's own servers rather than a third party's, and nothing is used to train them. Where a provider offers a setting that prevents my data being used for training, I use it.

Where I do not use it

• Your session notes, or anything drawn from our sessions.
• Assessment, formulation, risk work or safety planning.
• Any decision about whether to offer, continue or end counselling.
• Letters, reports or referrals about you.

I remain responsible for reading and checking anything an AI tool produces before I rely on it, and for keeping my own judgement and our working relationship central.

Worksheets in particular

A worksheet I give you is a piece of clinical material, so it gets more than a glance. I check it against my own training before you ever see it, I make sure it fits how we are working rather than dropping in something generic, and I am responsible for what it says.

There is a second check worth mentioning. These tools are trained on published material, so a drafted worksheet can come out closely resembling a copyrighted one from a published workbook or manual. I look for that too, and I do not pass off someone else's instrument as my own. Where I use a recognised measure such as the GAD-7 or PHQ-9, it is the published instrument, used as intended, and not something an AI has reconstructed.

The assistant on my website

There is one place where AI and your own information can meet, and I want to be straightforward about it. If you type something into the assistant on my website, what you type is sent to Anthropic so that it can produce a reply. That is you choosing to enter your own information — it is not me entering information about you.

The assistant is not a counsellor. It does not provide therapy, advice or emotional support, and it is not for use in a crisis. Conversations are not stored on my website: they exist only while the chat panel is open and are gone when you close it. They are not used to train AI models, and they form no part of your counselling record.

My Privacy Notice sets out the full data flow, and my Terms of Use set out the conditions of using the site and the assistant.

My workbook

I sell a self-help workbook to the public. It is a separate commercial activity from my counselling practice, and I want to be clear about how it was made and how it sits alongside the therapy I offer.

• The ideas and the content are mine. I decided what went into it and how it was structured, and I worked through more than a dozen drafts before it was finished. Where I used AI it was for wording and formatting — the job a good editor and a typesetter would do between them. I am responsible for every word of it.
• Nothing in it comes from any client, any session, or any of my notes.
• It is checked for clinical accuracy against my own training, and checked so that it does not reproduce material from published workbooks or manuals.
• It is a self-help resource. It is not therapy, it is not a substitute for therapy, and it is not a diagnostic or screening tool.
• It is not endorsed, approved or recognised by BACP. My registration tells you that I am a registered counsellor; it says nothing about this workbook.

I do not sell it to my counselling clients. Where it would be useful in our work together, I give it to you free of charge. You are never asked to buy anything from me.

Marketing and social media

I use AI to help draft posts and captions, mostly for Instagram. Two rules govern that. I read and approve everything before it goes out, and nothing I publish claims an outcome I could not stand behind — my client agreement says plainly that I cannot guarantee any particular result, and my advertising will not imply otherwise.

I do not use AI to write testimonials, reviews, or anything presented as another person's words, and I do not publish invented accounts of client experience.

Notes and anything drawn from our sessions

I do not use AI for your notes, or for any content derived from our sessions. My notes are written and kept by me. No material relating to you is entered into any AI tool. Tools now exist that could do this; that a tool could do something does not mean I should.

Consent, transparency and your right to say no

Because no identifying client information goes into an AI tool, I am not asking you to consent to your data being processed by one. What I owe you instead is honesty that AI plays any part at all, and a real choice about it.

Worksheets are the case where this matters most in practice. If I hand you a worksheet, it may well have been drafted with AI help — from a generic prompt, with no part of it drawn from anything you have told me. I would rather you knew that from this policy and from our agreement than wonder about it afterwards, and you are welcome to ask me about any material I give you.

• My use of AI is set out in the client agreement you sign before we begin, and I am happy to talk it through with you at any point.
• If you would prefer that I use no AI tools at all in connection with the administration of your counselling, tell me and I will work without them. It will make no difference to the counselling you receive, and you do not need to give a reason.
• If you change your mind in either direction, you can tell me at any time.

How I assess a tool before I use it

Before I use any AI tool, digital tool or online platform in my practice, I assess it and write that assessment down. I look at what information would touch it, who the provider is and where the data go, what could go wrong, what I do about each risk, whether I am competent to use it, and what I do if it fails. I keep those assessments and review them at least once a year, and sooner if a provider changes its terms or something goes wrong.

I also keep a fallback for anything I rely on, so that a tool failing or becoming unavailable does not disrupt your counselling.

What I will not use AI for

• Generating images, video or audio of clients.
• Persuading a client to change their fundamental beliefs, values or sense of self.
• Facilitating romantic or sexualised contact, or soliciting payment.
• Misrepresenting reality, or anything that damages the authenticity of the counselling relationship.
• Making or supporting any decision about whether to offer, continue or end your counselling.
• Writing testimonials, reviews, or anything presented as another person's words.
• Making claims about outcomes, or any claim I could not evidence if I were asked to.

If you use AI yourself

Many people use AI in their own lives, and some bring it into counselling — an answer a chatbot gave them, or a conversation they found helpful at three in the morning. I do not dismiss that. I am interested in what it gives you and what it costs you, and I stay alert to over-reliance, dependency, and the way these tools can reflect your own thinking back at you rather than challenge it.

If something you have taken from an AI tool seems as though it might be harmful, I will say so, and I will invite us to think it through together rather than simply telling you to stop. I would rather talk openly about what AI can and cannot offer, so that you can make your own informed choices about it.

Data protection

Any use of AI complies with the UK GDPR, my Privacy Notice, and my Data Protection and Information Security Policy — including having a lawful basis, keeping data to a minimum, and having a processor contract in place where a provider handles personal data on my behalf.

This policy also reflects core responsibility 2.1(e) of the BACP Ethical Framework for the Counselling Professions 2026, which requires me to assess the risk of any AI tool, digital tool or online platform before using it, to be competent to use it, to understand how the data are handled and stored, to be honest and transparent with people about my use of it and about the benefits and risks involved, to have mitigated the risks as far as I can, and not to hand my decision-making over to it.

Questions

If you have a question about anything here, or you would like me to work without AI tools, email me at jplcounselling@protonmail.com or call 07340 131355.