20,000 NHS scribes, CQC guidance, MHRA reports: this week in AI

If you spent last week focused on running your practice rather than monitoring healthcare AI news, here is what you missed. This fortnight produced a CQC guidance update that reframes the compliance question for any practice using AI tools, the largest NHS ambient scribing deal in history, two MHRA evidence reports that signal where regulation is heading, a finding from the UK's biggest health conference about why AI adoption stalls, and a practical procurement tool that most independent practice managers do not know exists. Each item has a direct implication for your decision-making. Here they are.

1. CQC June AI guidance: no new framework, but governance evidence is now mandatory

The Care Quality Commission published updated AI guidance in June 2026 that clarifies a question many practices have been asking: is there going to be a separate AI regulatory framework? The answer, for now, is no. CQC's position is that existing regulation already covers the use of AI in healthcare settings: the safe, effective, and well-led quality standards apply to AI tools in exactly the same way they apply to any other clinical or administrative tool.

What the guidance does specify, clearly, is what "well-led" looks like when AI is involved. CQC expects practices to show that AI tools support safe, person-centred care; that professional judgement sits behind every AI-assisted decision; and that the governance around AI use is sound and demonstrable. The word "demonstrable" is the operative word. Documentation, not intention.

The number: 24. CQC's draft revised assessment framework reduces the key lines of enquiry from 34 to 24. But the well-led section, where AI governance sits, is not among those being cut. Source: CQC framework pilot documentation, June 2026.

The so-what: if your practice uses any AI tool that touches patient data or informs a clinical interaction, you need four things on file before October 2026 when the framework pilot evaluation concludes. A completed Data Protection Impact Assessment. A signed Data Processing Agreement with your AI vendor. Evidence of staff training. A patient consent process for tools used in the consultation room. Most independent practices currently have none of these. The CQC AI governance checklist we published last week walks through each one in detail.

2. South West London NHS trusts sign the largest AI scribe deal in NHS history: 20,000 clinicians, mandatory patient consent

In April 2026, the South West London Acute Provider Collaborative announced a partnership with Lyrebird Health to deploy ambient voice AI scribing technology across four NHS trusts: St George's University Hospitals, Epsom and St Helier University Hospitals, Croydon Health Services, and Kingston and Richmond NHS Foundation Trust. The initial phase targets 10,000 clinicians, scaling to 20,000 over the programme lifetime. It is the largest single NHS ambient scribing deployment to date.

The deal followed NHS England's publication of updated national guidance on AI-enabled ambient scribing products in April 2026. The guidance establishes a firm consent requirement: clinicians must tell patients at the start of each interaction if AI scribing technology is being used, and patients must be given the opportunity to object. If a patient objects, the tool cannot be used for that session.

The number: 20,000. Clinicians in the South West London rollout. Source: South West London Acute Provider Collaborative and Lyrebird Health announcement, April 2026.

The implication for independent practices is not that you are in this rollout (you are not). It is that the consent standard being embedded in NHS practice will become the de facto expectation for all clinical AI use. GDPR and HCPC/GDC professional standards already point the same direction. If your practice is using or evaluating any AI tool present in the consultation room, a documented patient disclosure process is no longer optional. Build it now, before a complaint or inspection makes it urgent.

3. MHRA publishes two landmark evidence reports on AI regulation

On 11 June 2026, the Medicines and Healthcare products Regulatory Agency published the findings from its National Commission into the Regulation of AI in Healthcare: two reports drawing on evidence from patients, the public, healthcare professionals, industry, and system partners gathered between December 2025 and February 2026.

The central finding is that the current regulatory framework, designed for static medical devices with a one-time approval process, is not well suited to AI systems that update and adapt over time. Stakeholders across the evidence base called for a risk-based, proportionate approach that can accommodate iterative development, while maintaining patient safety as the primary standard. The Commission's formal recommendations to the MHRA are expected later in summer 2026.

The number: 2. Evidence reports published. Commission formal recommendations: due summer 2026. Source: MHRA, 11 June 2026.

The so-what for independent practices is about timing. AI tools that currently sit in a regulatory grey area (ambient scribing tools, AI-assisted scheduling, AI outcome tracking, AI treatment planning features) are going to face clearer classification requirements once the MHRA recommendations land and any resulting rule changes begin. Practices that have already done a proper governance review of their AI tools will be well-positioned. Those that have not will need to catch up on someone else's timeline. If you are evaluating a tool now, asking your vendor "how will your classification change under the new MHRA framework?" is a reasonable question. Their answer will tell you something useful about their governance maturity.

4. NHS ConfedExpo 2026: the barrier is trust, not resistance

The NHS Confederation's annual expo, held in Manchester on 10–11 June 2026 and attended by thousands of health and care leaders, put AI at the centre of its programme this year. The findings from the event are worth reading carefully if you are a practice manager who has struggled to get clinical staff on board with AI adoption.

The headline finding: there is no widespread resistance to AI among NHS clinicians. What there is, consistently across every panel and discussion tracked at the conference, is a set of reasonable questions about clinical safety, information governance, data sovereignty, reliability, and appropriate human oversight. These are not objections to be overcome. They are due diligence questions that require substantive answers.

The number: 0. Widespread resistance to AI observed at NHS ConfedExpo 2026. The conversations were about evidence and governance, not whether AI should be used at all. Source: NHS ConfedExpo 2026 reporting, Medical Device Network / Trustmarque post-event analysis, June 2026.

If your clinical staff are asking questions about AI tools you are proposing to implement, they are doing exactly what a CQC inspector would want them to do. The correct response is not a sales pitch. It is evidence: clinical safety documentation from the vendor, your own DPIA, a clear answer to "what happens to patient data," and a consent process. Practices that have done the governance work find that clinical buy-in follows. Those that lead with the technology and skip the governance find that it does not. That pattern held across 30,000 NHS workers in the Copilot trial and it will hold in your practice too.

5. The NHS AI scribe registry: 20-plus approved suppliers and how to use it for independent practice procurement

NHS England launched a self-certified supplier registry for AI ambient scribing products in January 2026. The registry now lists more than 20 suppliers, including Heidi Health, Lyrebird Health, Accurx, TORTUS, Microsoft Dragon Ambient eXperience, eConsult, and others. To appear on the registry, suppliers must self-certify compliance with NHS standards covering clinical safety (DCB0129 and DCB0160 frameworks), technology, and data protection.

The registry was designed for NHS procurement. Most independent practice managers do not know it exists.

The number: 20-plus. Suppliers currently listed on the NHS England AI ambient scribing registry. Source: NHS England / Digital Health, April 2026.

The practical use for an independent practice is straightforward. If you are evaluating ambient scribing tools and a supplier appears on the NHS registry, that supplier has at minimum self-certified against clinical safety and data protection standards that NHS procurement requires. It is not a guarantee of fit for your practice or specialty. It does not replace your own vendor due diligence. But it is a starting point that a Google search cannot give you, and it narrows the field considerably. The registry is searchable at england.nhs.uk. If you are comparing tools and one vendor is not on the registry, that is a question worth asking them directly: why not?

What this week adds up to

Taken together, these five developments tell a consistent story. The NHS is deploying AI at scale. Regulators are clarifying expectations without creating new frameworks. Clinicians are ready, but they need governance evidence, not product demos. And the tools that practice managers need to make good procurement decisions are available, but not well signposted.

The common denominator in all five stories is governance documentation. A DPIA. A DPA with your vendor. A patient consent process. Staff training records. Our CARE Framework data puts the average independent practice at 41/100 on AI readiness, with governance the lowest-scoring dimension every time. That gap between where most practices are and where the regulatory environment expects them to be is closing faster than most practice managers currently expect.

If you want to understand exactly where your practice stands before CQC's pilot evaluation completes in November 2026, the AI Opportunity and Growth Assessment takes two weeks and covers the full governance gap alongside tool evaluation and ROI modelling. Or book a 20-minute discovery call to talk through your specific situation first.