HCPC's consent standard is binding. BPS's AI guidance isn't.

Ask a psychology practice manager whether adding an AI note-taking tool requires fresh client consent, and most will say yes, then struggle to name which rule actually says so. The honest answer is that two different bodies have something to say about it: one whose standard is legally enforceable and does not mention AI at all, and one whose AI-specific guidance is detailed and useful but carries no statutory weight on its own. Mixing the two up is the most common gap we see in psychology practice consent paperwork, and it costs nothing to fix once the split is clear.

This matters more for psychology than for most specialties, because consent sits at the centre of the therapeutic relationship rather than at its edge. Getting it wrong is not a paperwork problem waiting to be found at inspection. It is a clinical governance problem with a direct route to a fitness-to-practise complaint. So it is worth being precise about which document actually binds you, and which one is advisory.

The one consent standard that is actually binding

The Health and Care Professions Council regulates 15 health and care professions by statute, including practitioner psychologists, and keeps a public register of everyone who meets its standards (HCPC, Standards of Conduct, Performance and Ethics, effective 1 September 2024). Standard 1.4 of that document states: "You must make sure that you have valid consent, which is voluntary and informed, from service users who have capacity to make the decision or other appropriate authority before you provide care, treatment or other services."

That single sentence is the whole of HCPC's binding consent requirement. It is not AI guidance. It is a general standard that has applied to every registered practitioner psychologist since September 2024, and it is enforceable: HCPC can and does act on fitness-to-practise concerns where consent was not genuinely voluntary and informed, whether or not an AI tool was involved in the service.

So what for you: if a complaint or a safeguarding review ever asks whether a client consented to an AI-assisted session, Standard 1.4, not any AI-specific document, is the rule your answer will be measured against.

What HCPC's own standard does not say about AI

Read the full 2024 standards document and AI does not appear once. HCPC has been active on AI elsewhere: in February 2026 it joined four other regulators, the General Osteopathic Council, General Optical Council, Royal College of Veterinary Surgeons and General Pharmaceutical Council, in a joint statement on AI (HCPC, joint statement on AI in health and care professional education, 9 February 2026). That statement is specifically about AI in education and training for future registrants, covering academic integrity and AI literacy on approved programmes. It says nothing about consent for AI tools used with existing clients in practice.

There is, as of this writing, no HCPC-specific clinical guidance interpreting how Standard 1.4 applies to AI. That gap is exactly where the confusion practices report starts, and it is unlikely to close quickly given how HCPC's other regulatory workstreams have moved.

So what for you: do not wait for HCPC to publish an AI-specific consent standard before acting. Standard 1.4's general wording already covers any care, treatment or other service, which includes AI-assisted ones. The obligation exists now, whether or not the word "AI" appears anywhere in HCPC's text.

BPS's AI guidance says more, but it is not the law

The British Psychological Society is a professional membership body, not the statutory regulator for psychologists in the UK. That distinction matters here specifically: BPS membership and its ethical code carry professional weight, but not the same enforcement route as HCPC registration.

In April 2026, BPS endorsed the Global Psychology Alliance's Top 10 Principles for Understanding Artificial Intelligence in Psychology, a guidance document covering ethics, professional competence, and access, bias and inclusion. On consent specifically, the principle is direct: psychologists have an ethical obligation to obtain informed consent by clearly communicating the purpose, application, and potential benefits and risks of any AI tool involved in a client's care, in language the client can actually understand, not a vendor's technical description.

This is more specific and more useful than anything HCPC has published on AI consent. It is also, on its own, non-statutory. A psychologist who ignores it faces no direct regulatory sanction for that alone. Where it bites is indirectly: if the AI disclosure a practice gives a client is vague or technical enough that a regulator would not call it genuinely informed, that failure lands back on HCPC's Standard 1.4, not on the BPS guidance itself.

So what for you: treat HCPC's Standard 1.4 as the reason consent is compulsory, and the BPS-endorsed principles as the working answer to what that consent conversation should actually contain.

Consent is not a one-off form

BPS's own Code of Ethics and Conduct treats informed consent as a dynamic process rather than a single signature. Consent should be reviewed when there is a substantive change in the intervention being offered, or when a psychologist has reason to think a client's original consent may no longer hold (British Psychological Society, Code of Ethics and Conduct, 2021).

Neither BPS nor HCPC has published a document that says explicitly "adding a new AI tool counts as a substantive change requiring re-consent." That is our reading of two existing principles read together, not a quoted rule, and it should be treated as a reasonable interpretation rather than a settled regulatory position. But the interpretation is a defensible one: moving a client from handwritten session notes to an AI transcription tool, or adding an AI-assisted intake or triage step, changes how their information is captured and processed in a way that a client who consented before the tool existed was never actually told about.

So what for you: build a checkpoint into your practice's process so that whenever a new AI tool is introduced, or an existing one changes materially, current clients are told and given a genuine chance to ask questions or object, rather than assuming a signature taken at the start of therapy still covers it.

A scope caveat worth stating plainly

Everything above applies specifically to HCPC-registered practitioner psychologists. Not everyone providing psychological therapy in the UK holds that registration. Some counsellors and psychotherapists are members of voluntary bodies such as BACP or UKCP instead, whose codes carry a different, non-statutory status. If your practice employs a mixed team, the binding-versus-advisory split described in this article does not apply uniformly across everyone in the building.

So what for you: map which staff are HCPC statutory registrants and which are governed by a voluntary body's code only, before assuming every clinician in your practice owes clients the same legal consent obligation.

What this means for your practice

Four actions, in order of priority.

First: write a specific AI disclosure statement for each AI tool in use, covering its purpose, what it does with client information, and its main benefits and risks, in plain language rather than a vendor's marketing description. This is the content BPS's endorsed principles call for, and the evidence HCPC would look for under Standard 1.4.

Second: treat the introduction of any new AI tool, or a material change to an existing one, as a consent-review trigger for current clients, not just something covered by new-client intake paperwork.

Third: document consent conversations specifically, not just a checkbox marked "consent obtained," so you can evidence that consent was voluntary and informed if a complaint or review ever asks. This connects directly to the data protection side of the same problem: for the Special Category data and DPIA rules that sit alongside consent, see AI note tools in therapy: the compliance rules that now apply.

Fourth: map staff registration status across a mixed team so the right consent obligation is applied to the right clinician, rather than assuming HCPC's Standard 1.4 covers everyone delivering therapy in your practice. Consent sits alongside, not instead of, the safeguarding obligations CQC expects around AI use; if you have not already closed that gap, see CQC names 5 regulations for AI. Safeguarding isn't one.

None of this requires slowing down or reversing an AI tool that is already working well for your practice. It requires a short, specific piece of paperwork and a habit of re-checking it, rather than a fresh technology purchase. If you want an independent view of where your practice's consent documentation actually stands against these requirements, the AI Opportunity & Growth Assessment™ reviews consent and governance as part of its standard scope. You can also book a 20-minute call to talk through where your practice sits before committing to anything further.

The single most important thing to take from this

HCPC's Standard 1.4 was written before generative AI existed in its current form, and it does not need updating to apply to it. It already covers any care, treatment or other service you provide, AI-assisted or not, and it is the standard a regulator will actually use if your consent process is ever questioned. BPS's AI-specific guidance is not a substitute for that standard. It is the best available answer to what a compliant consent conversation should say. Practices that treat the two as interchangeable are the ones most likely to discover, at the worst possible moment, that a signature taken months ago never covered the tool they are using today.