66,695. That is how many children and young people in England were on a waiting list for speech and language therapy as of April 2026, according to a written answer given in Parliament by the Department of Health and Social Care on 1 July 2026 (Department of Health and Social Care, written answer UIN 11799, 1 July 2026). NHS England has set a target for 80% of community health services activity, including speech and language therapy, to happen within 18 weeks by 2028/29, and for integrated care boards to eliminate 52-week waits during 2026/27. Neither target has been met yet.

Families who cannot wait turn to independent practice. The Association of Speech and Language Therapists in Independent Practice (ASLTIP) now represents more than 1,800 members across the UK, every one of whom must also be registered with the Health and Care Professions Council (HCPC) and the Royal College of Speech and Language Therapists (RCSLT) (ASLTIP, 2026). RCSLT's own 2025 vacancy survey found 14% of UK SLT posts unfilled, rising to 17% in England across both the NHS and independent sectors, a shortage that pushes more of the work that does get delivered onto fewer clinicians (RCSLT, Workforce and Vacancy Survey, 2025). A growing share of that work happens over video rather than in a clinic room, and a UK survey of SLTs treating people with post-stroke aphasia found the large majority planned to keep offering telehealth alongside in-person sessions well after pandemic restrictions ended (published in PMC, checked 2026).

The guidance most solo SLTs have not read yet

On 7 May 2026, RCSLT published "Artificial Intelligence (AI): principles for safe and ethical practice in speech and language therapy," a 12-principle framework developed in direct response to member demand (RCSLT, 7 May 2026). It goes further than comparable guidance from allied health bodies this site has reviewed elsewhere. In the US, the American Speech-Language-Hearing Association (ASHA) gives certified professionals only a general duty under its Code of Ethics to "consider and evaluate any technology used in practice," and states plainly that there is limited legislative or regulatory oversight of generative AI in speech-language pathology at present. ASHA has not published an AI-specific consent standard (ASHA, "Generative Artificial Intelligence for Clinicians," asha.org, checked 2026). RCSLT's document runs to 12 numbered principles covering transparency, need and purpose, safety, effectiveness, equity and bias, confidentiality, information governance, sustainability, consent and choice, clinical accountability, scope of practice, and organisational governance, each with specific expectations rather than a general instruction to be careful.

So what for you: if your continuing professional development on AI has leaned on US-sourced webinars or blogs, treat that as background only. Read RCSLT's primary document once in full before you touch a new tool, because the standard you are actually held to is more specific than most secondary summaries suggest.

Two principles that apply directly to a teletherapy session

Two of the 12 principles describe, almost exactly, what a video appointment produces. On confidentiality, RCSLT states that identifiable client information, "including names, case details, speech samples, transcripts, audio recordings, video recordings, assessment data or personal details," must not be entered into an AI tool unless that system has been formally approved for secure clinical use, and flags this as "particularly important" when using publicly available or consumer AI systems (RCSLT, May 2026). On information governance, the guidance requires AI use to comply with GDPR and data protection policy, "particularly when processing audio, video, and transcribed language data." That is a precise description of a recorded teletherapy call.

A third principle, consent and choice, requires SLTs to seek informed consent wherever AI use influences assessment, clinical decision making, therapeutic interaction, or the processing of personal data, and states that clients must have the option of support directly from an SLT without being required to use an AI-supported tool as a substitute.

So what for you: before your next video session where you plan to run an AI note-taker in the background, get documented consent for that specifically, separate from consent for the video call itself. RCSLT's own telehealth guidance already sets that bar for the call; the AI principles extend it to whatever tool is listening in.

A tension nobody has resolved yet: local storage advice meets cloud-based AI tools

RCSLT's telehealth guidance, which predates the AI principles, already covers recording video sessions. It states that consent can be given verbally or in writing but must be documented in your clinical notes, and that where a recording is made, it should be saved to a secure local drive, meaning a drive on your own computer, rather than uploaded to cloud storage (RCSLT, Telehealth guidance, members area, checked 2026).

That instruction sits awkwardly next to nearly every commercial AI note-taking and transcription tool on the market, almost all of which are cloud-hosted by design: audio goes to a server you do not control, gets processed, and a transcript or summary comes back. RCSLT's newer AI principles do not resolve this directly. They require that any tool handling identifiable data be "formally approved for secure clinical use," which for a sole practitioner with no IT department or information governance team means doing that vetting yourself, checking a vendor's data processing agreement, where servers are hosted, whether recordings are used to train the underlying model, and how long data is retained, before a single session gets recorded through it.

So what for you: treat storage location and retention period as a pass or fail question when you evaluate any AI note-taking tool for teletherapy, not a detail to check later. If a vendor cannot give you a written answer on where audio and video are stored and for how long, do not use it for identifiable client sessions, regardless of how good the transcription quality is.

Where AI genuinely saves a solo SLT time

RCSLT's own principles document names where it sees "clear and meaningful benefit" from AI: reduced administrative burden, improved access to information, better development and adaptation of therapy resources, and improved documentation and communication. For a sole practitioner combining direct sessions with report writing, letter drafting, and resource preparation, most of the reclaimable time sits in that second half of the list, not the therapy itself.

WriteUpp, used by more than 50,000 clinicians across the UK and Canada, is one of the few UK practice management platforms built with SLT-specific note formats (voice assessment, case history, dysphagia) alongside built-in video consultations, meaning the video call and the documentation tool can run through the same vendor rather than two separate ones (WriteUpp, product pages, checked 2026). Its AI Medical Scribe add-on is priced at £23.95 per user per month with four free trial hours (WriteUpp, via IBTimes UK and company press materials, 2026), a vendor-published figure worth testing against your own note-writing time rather than taking as a guaranteed saving. What it is not is a substitute for checking that add-on's own data handling against RCSLT's principles before you switch it on for a client session; using the same vendor for the call and the notes reduces the number of separate data flows you need to vet, it does not remove the need to vet them.

So what for you: a general-purpose transcription tool built for business meetings, not clinical use, is the wrong choice for a client session regardless of price, because it almost certainly fails the confidentiality principle outright. Spend the time comparing tools that are actually built for regulated clinical documentation instead.

Before you switch anything on: a same-day checklist

Treated literally, RCSLT's principles set out concrete actions rather than a mood. Update your privacy notice to name any AI tool you use and explain, in terms a client or parent can understand, what it does with their information. Confirm your Information Commissioner's Office data protection fee registration is current, since AI tools processing client data do not change this requirement. Ask any AI vendor, in writing, where audio, video and transcript data are stored, whether they are used to train the underlying model, and how long they are retained. Build a short, spoken consent script for teletherapy sessions that names the AI tool specifically, not just the video platform, and document that consent in your notes every time. Ask your professional indemnity insurer whether they require a specific declaration about AI use in independent practice.

None of this requires new equipment or a large budget. Most of it is a single afternoon's work, and it is the same checklist regardless of which specific tool you end up choosing.

So what for you: build this into how you evaluate every new AI tool going forward, not as a one-off task tied to your first purchase. The waiting list pressure driving more families toward independent SLTs is not going away by 2028, and neither is the expectation that you can answer a parent's question about what happens to their child's recorded session, on the spot, in plain language.

The clearest starting move is to read RCSLT's AI principles in full alongside your telehealth guidance, then check any tool you are considering against both before you record a single session with it. If you want an independent view of where AI actually saves time in your specific practice, scored against the same governance questions covered here, the AI Opportunity and Growth Assessment does that in a single report, with no vendor relationships behind the recommendation. The starting point is a free 20-minute discovery call.

Not sure whether your teletherapy AI risk sits in consent, storage, or vendor vetting? A short call is usually enough to tell. Book a 20-minute call.

The Clinical AI Briefing

One practical AI insight for healthcare practices every week. No hype. Evidence and outcomes only.

Related: AI HIPAA rule delayed to 2027. Safeguarding duty isn't. · The $2,480 threshold: what AI billing tools actually fix for solo OT and SLT practices · HCPC's consent standard is binding. BPS's AI guidance isn't.