For BACP members and UK counsellors · checked 22 August 2026
The BACP Ethical Framework 2026: what changes, when, and what to do before November
The new framework becomes mandatory at midday on 3 November 2026. Here are the key dates, the three big changes, and the new clause on AI tools in plain English. Then a free 90-minute course with a certificate on successful completion, if you want to go deeper.
Free 90-minute course + completion certificate. Launching late September. One email to sign up.
The short version
There is no separate "BACP AI framework". What people are talking about is the Ethical Framework for the Counselling Professions 2026: BACP's full revision of the framework every member signs up to. It replaces the 2018 version, and for the first time it says something specific about AI and digital tools.
BACP published it on 4 August 2026. It becomes mandatory at midday on Tuesday 3 November 2026. Until then the 2018 framework still applies, and the three months in between are for reading, asking questions and getting your practice ready.
Three things change in substance. The framework now covers how you work ethically with everyone in your professional life, not only clients. It asks you to be able to give a rationale for your ethical decisions, in a form you can use in supervision, CPD planning and, if it ever comes to it, a professional conduct process. And it adds a core responsibilities section that includes, for the first time, the use of AI and digital technology.
The rest of this page takes those in turn, with the AI clause spelled out line by line, because that is the part most people have questions about.
Key dates to November
- 4 August 2026DoneBACP publishes the 2026 Ethical Framework (PDF and Word versions on bacp.co.uk) and opens a three-month familiarisation period.
- August to OctoberNowRead it, compare it with the 2018 version, and bring questions to supervision. BACP is releasing supporting material through this period: Ethics Essentials videos, a glossary and FAQs, and updated Good Practice in Action resources reorganised into three categories.
- Late SeptemberOur partOur free 90-minute course on the 2026 framework and the AI clause opens to the waitlist. Completion certificate included. Details below.
- 1 November 2026In the PDFThe date printed inside the framework itself: "takes effect from 1 November 2026".
- 3 November 2026MiddayThe date on BACP's website: the 2026 framework replaces the 2018 framework and becomes mandatory for all members. Safe reading: have everything in place by the end of October.
What changes, in three parts
1. Ethical practice across all your professional relationships
The 2018 framework was written largely around the practitioner and the client. The 2026 version opens that out: the same ethical commitments now apply to how you work with colleagues, supervisors, supervisees, trainees, employers, organisations and the public. If you supervise, teach, or sit on a team, the framework now speaks to that part of your work directly.
2. A rationale for your ethical decisions
This is the change that runs through everything else. The framework asks members to be able to explain their reasoning for the decisions they make and the actions they take, and it sets out the questions a rationale should answer: what the ethical issue was, how you engaged with it, what factors inside and outside the situation shaped it, what you decided, what happened, and how you reviewed it afterwards.
In practice that means "why did you do that" becomes a question you should be able to answer in writing, not just in the moment. Supervision is the natural home for it. It also turns out to be the key to the AI clause, as you will see.
3. Core responsibilities, including AI and digital technology
The new framework has a core responsibilities section with four strands: building appropriate relationships; keeping yourself, colleagues and clients safe; understanding and acting on your legal and professional responsibilities; and delivering services to high professional standards. The AI and digital tools wording sits in that last strand, in clause 2.1(e). It is the first time the framework has named AI at all.
The AI clause in plain English
Clause 2.1(e) asks you to assess the risk of any AI tool, digital tool or online platform before you use it, and then to be able to demonstrate five things. Here they are, with the framework's own words and what each one means on a Tuesday afternoon.
One thing to hold onto as you read: the clause is written around you, the member. It does not name any product, it does not ban anything, and it does not create a list of approved tools. There is no BACP scheme that approves or certifies AI tools. Any claim that a tool is "BACP-compliant" or "BACP-approved" describes something that does not exist.
-
You are competent to use the tool
"are competent to use the tool or platform" (2.1(e)(i), PDF)
Not the engineering, the practice. You know what goes in, what comes out, where it ends up, how to correct it and how to stop it. A test run on yourself or a colleague before a real client is the simplest evidence there is.
-
You understand how data are handled and stored
"understand how data are handled and stored, and any risks to confidentiality and that we have mitigated any associated risks as far as possible" (2.1(e)(ii), PDF)
Where the data goes (a country, not "the cloud"), who else touches it, whether it is ever used to train AI, how long it is kept, and how you delete it. These are questions for the vendor, and a vendor who cannot answer them has answered them. For general-purpose AI assistants, the same questions apply, and the answers are often the reason they are not suitable for identifiable client material.
-
You are open with clients about using it
"are honest and transparent with people about our use of AI, digital tools or online platforms, and the benefits and risks involved with their use" (2.1(e)(iii), PDF)
A sentence you can say out loud in a first session, and something written the client can read again later: what the tool does, what it does not do, where their information is, and that they can say no.
-
You have informed consent before any of their data goes in
"have the informed consent of those using our services before inputting any of their personal data into AI or digital tools" (2.1(e)(iv), PDF)
This is the one most likely to catch people out, so it is worth being literal about it. "Before inputting" means before. Before a name, a session summary or a recording goes into any AI tool: a notes app, a dictation tool, a chatbot you use to tidy up a letter. Not retrospectively, and not implied by the fact that the client kept coming. In everyday terms: a separate, explicit yes, recorded somewhere, given before the first time you use the tool for that client. For recorded sessions, a fresh yes each time recording starts. A client can decline and still work with you. If you work with children, young people or adults who may lack capacity, think about who consents and how. And if you are already using an AI tool with existing clients, that conversation needs to happen before November, not after.
-
The decisions stay yours
"have not deferred or outsourced our decision-making to AI tools. We remain responsible for critically evaluating AI output" (2.1(e)(v), PDF)
A draft note is a draft. A suggested formulation is a suggestion. You read every AI output before it becomes part of a record or a decision, you change what is wrong, and the assessment is yours. The shortest of the five and the one that matters most if anyone ever reads your notes.
Two neighbouring clauses worth reading with it
Privacy notice and confidentiality (3.1(c) and (d)). Clients should have a clear, accessible privacy notice covering how their personal data are collected, used, stored and protected, and they should be made aware of, and agree to, any foreseeable limits to confidentiality, including the use of digital storage systems, platforms or tools that may monitor or collect data. If you use an AI tool, your privacy notice needs to say so.
Records (3.2). Records must be factual, adequate and relevant, stored securely, and meet the data protection requirements of the country where the record is stored. So you need to know the country. "EU" is a better answer than "the cloud". A named country is better still.
Put the three together with the new rationale requirement and the shape becomes clear: for every AI or digital tool you use, you should be able to say what you checked, what you tell clients, how they consented, and why you chose it. That is an afternoon's work once, and a few minutes per new client afterwards.
90 minutes, certificate on completion, launching late September.
What to do before November
A plain checklist. Most of it is writing things down that you may already do.
- Read the framework itself, not a summary of it. It is shorter than you expect. Read the 2018 one again alongside it if you want to see what moved.
- List every digital and AI tool that touches client information. Notes and transcription tools, dictation, scheduling, video platforms, cloud storage, and any general AI assistant you use for letters or admin. The general ones are the ones people forget.
- For each tool, answer the five points and keep the answers with a date: competence, data handling, transparency, consent, decisions. Where the vendor cannot answer, write that down too. "Asked on [date], no answer yet" is a legitimate entry.
- Update your privacy notice so it names the digital tools that may collect or store client data (3.1).
- Add consent to the tool to your client agreement, as a separate, explicit yes. For existing clients who are already on an AI-assisted tool, have the conversation before November.
- Write your rationale for each tool you keep: what it does, what you checked, the risks and how you reduce them, what clients are told, and when you will review it. Five lines is enough.
- Take it to supervision and put a review date in the diary, a year out or sooner if the vendor changes its terms.
- Put 3 November in the diary and, if you are a member of more than one body, check what the others say. NCPS and UKCP have published their own views on AI and they are not identical.
Free course · launching late September
The 2026 framework and AI, in 90 minutes
A free course that walks through the new framework, takes the AI clause point by point, and leaves you with your own written rationale, consent wording and client explanation by the end. Self-paced, online.
From the team behind Joy, a Swedish company building tools for therapists. Not affiliated with BACP.
A note from Charlotta
Hi, I'm Charlotta, one of the co-founders of Joy. I have spent more than a decade building healthcare technology, on both sides of the table: running companies that deliver care, and building the tools they run on.
We wrote this page because the 2026 framework asks something reasonable of you, and the information about it is scattered. The course exists for the same reason. It is free, it will stay free, and it is not a sales pitch for anything. If it leaves you with a clear rationale and a consent wording you are happy with, it has done its job.
If you think we have read a clause wrong, or something here goes out of date, email me at charlotta@joy.day. It comes straight to me, and I read everything.
Charlotta
Co-founder, Joy
Quick answers
When does the BACP Ethical Framework 2026 come into effect?
BACP's website says the 2026 framework becomes mandatory at midday on Tuesday 3 November 2026. The framework document itself says it takes effect from 1 November 2026. Until then, the 2018 framework still applies. Either way, aim to have your practice ready by the end of October.
Is there a separate BACP AI framework?
No. The AI wording is part of the full 2026 Ethical Framework, in clause 2.1(e) of the core responsibilities section. There is no standalone AI policy, no list of approved tools, and no BACP scheme that certifies or approves products.
What does the new BACP framework say about AI?
Clause 2.1(e) asks members to assess the risk of any AI tool, digital tool or online platform before using it, and to be able to show five things: that they are competent to use it, that they understand how data are handled and stored, that they are open with clients about using it, that they have informed consent before inputting any client personal data, and that they have not handed their decision-making to the tool.
Do I need a client's consent before using AI for their session notes?
Under the 2026 framework, yes. Clause 2.1(e)(iv) asks for informed consent before inputting any of a client's personal data into AI or digital tools, and a session note with their name in it is personal data. The consent should come before the first use, not after, and should be separate from a general agreement to therapy.
Does the 2026 framework ban AI tools?
No. It does not ban any tool or name any product. It sets out what you need to be able to demonstrate if you choose to use one. A practitioner who uses no digital tools at all still needs to meet the rest of the framework, and a practitioner who uses several can meet it with a few written records.
Is the free course accredited by BACP?
No, and we do not claim it is. BACP does not accredit or endorse this course or this page. The course is made by the team behind Joy, it is free, it comes with a completion certificate, and it counts towards your self-directed CPD in the same way as any other reading or learning you record yourself.
Want the 90-minute version? Join the waitlist and you will hear from us when the course opens.
Checked on 22 August 2026 against the Ethical Framework for the Counselling Professions 2026 (PDF) and BACP's Preparing for the 2026 Ethical Framework page. Short quotations from the framework are reproduced for the purpose of explanation; the framework is © 2026 British Association for Counselling and Psychotherapy, and the full text is theirs to read at the links above. This page and the course are made by Joy Solutions AB and are not affiliated with, endorsed by or approved by BACP. This is general information about a professional framework, not legal or ethical advice for your situation. Spotted something out of date? Tell us at hello@joy.day and we will fix it.