49% NHS AI trust gap, ICO's binding AI code: this week in AI

The NHS spent last week rolling AI tools into more hospitals. This week the rest of the system caught up with the questions that rollout raised: whether patients actually want the AI features on offer, whether the regulator meant to police AI and patient data now has legal teeth, and how far ahead of independent practice the wider AI market is moving. None of the five stories below need action from you this week. All five change what you should expect over the next twelve months.

1. Ambient voice technology reaches four more London trusts

Tens of thousands of NHS staff across south-west London are the latest to get ambient voice technology (AVT), AI tools that listen to a consultation and generate the clinical note, as four trusts, St George's, Epsom and St Helier, Croydon, and Kingston and Richmond, join the rollout. Source: NHS England and Digital Health, July 2026. The rollout sits inside the £10 billion technology programme confirmed on 4 July, and NHS guidance now explicitly favours AVT that writes directly into a trust's electronic patient record over standalone note-taking apps that require a separate copy-paste step.

The number: 4 trusts. That is the count this week, on top of the London trusts already running AVT and the national scaling plan announced two weeks ago. The detail that matters more than the count is the integrated-over-standalone preference. NHS England is not just buying more scribes, it is telling every trust which architecture to buy.

The so-what: independent practices weighing scribe options are facing the same trade-off the NHS just settled on a national scale. A scribe that writes straight into your practice management system beats one that hands you a transcript to paste in yourself, which is exactly the calculation we ran for a physiotherapy practice this week in our bundled-versus-standalone AI scribe decision. If the NHS's own procurement guidance is converging on integration over standalone tools, that is a reasonable steer for any practice choosing between the two.

2. The sector welcomes the money, questions the plan, and patients are still not sold

Reaction to the £10 billion announcement has been warm but pointed. Tim Horton, deputy director of policy at the Health Foundation, called the funding a positive step, then named the missing piece: a long-term strategy to guide how AI gets used across the health system, without which the NHS risks piecemeal adoption. The King's Fund separately flagged that evidence on achievable AI productivity gains remains limited, and that digital maturity still varies hugely between trusts. Source: Health Foundation press release, July 2026.

Patients are just as cautious. Health Foundation polling found only 49% of the public would use the proposed AI-powered "Doctor in Your Pocket" NHS App feature for non-urgent care, with 32% saying outright they would not. Source: Health Foundation polling, July 2026.

The number: 49%. That is barely half of the public willing to use a patient-facing AI feature the NHS is actively promoting, from an organisation patients already trust with their care.

The so-what: the same scepticism applies to any independent practice deploying patient-facing AI, whether that is an AI receptionist, a chatbot intake form, or automated recall messaging. Disclosure and informed consent, covered in detail this week for psychology practices in where BPS and HCPC guidance actually stand on AI consent, are not only a compliance requirement. They are a large part of whether patients use the tool at all. A practice that explains what the AI does and why, clearly and upfront, converts more of that scepticism into use than one that quietly switches a system on.

3. The ICO's AI code of practice stops being a promise and becomes a legal deadline

Last week's roundup noted that the ICO's promised AI and automated decision-making code of practice was still an undated draft. That has now changed in an important way. Regulations, UK Statutory Instrument 2026/425, came into force on 12 May 2026, placing the Information Commissioner under a statutory duty to produce a binding code of practice on AI and automated decision-making, with specific provisions covering children's personal data. Consultation on the draft guidance closed 29 May 2026, non-binding final guidance is due in summer 2026, and the full statutory code is expected to follow in 2027. Source: UK Statutory Instrument 2026/425; Information Commissioner's Office, AI and biometrics strategy update, 2026.

The number: 12 May 2026. That is when the legal requirement to write the code took effect, even though the code itself is still roughly a year away.

The so-what: this is not a reason to wait. The ICO's existing guidance on fairness and the treatment of vulnerable groups already applies today, the same gap we found in CQC's own safeguarding coverage in last week's look at CQC's five named AI regulations. Any practice using AI to inform decisions about patients, triage, prioritisation, risk flagging, should treat a DPIA against current ICO guidance as the baseline, and expect the 2027 code to formalise rather than replace it.

4. The US hands 2,000 public health workers free enterprise AI

The Coalition for Health AI launched PULSE, the Public health Use case and Learning Scaling Engine, on 16 July 2026. The initiative gives up to 2,000 public health practitioners at state, tribal, local and territorial agencies free enterprise AI licences, donated by Anthropic and OpenAI, to pilot five defined generative AI use cases in structured communities of practice. Source: Coalition for Health AI, 16 July 2026.

The number: 2,000. That is the scale two of the largest AI labs are now willing to fund directly inside public-sector healthcare, not as a paid pilot but as a donation.

The so-what: this initiative is scoped to public health agencies, not private independent practices, so there is nothing to sign up for here. But it is a clear signal of direction. When major AI vendors start donating enterprise access to build adoption evidence in healthcare, that evidence, positive or negative, tends to shape what gets marketed and priced for private practices within a year or two. Worth watching, not acting on.

5. The FDA clears an AI tool that reads heart disease from a routine ECG

Pathway Labs' EchoNext received FDA clearance on 1 July 2026 for six indications drawn from a standard 12-lead ECG, including right- and left-sided heart failure, valve disease, severe hypertrophy consistent with infiltrative cardiomyopathy, and pulmonary hypertension. These are conditions that normally require echocardiography or more invasive testing to confirm. Source: Pathway Labs, 1 July 2026.

The number: 6. That is how many distinct cardiac conditions one AI model is now cleared to flag from a test most practices already run routinely.

The so-what: cardiology sits outside every specialty this site covers, so there is nothing to act on here. What is worth noting is the pattern the FDA keeps approving: AI reading more signal out of data a practice already collects, rather than requiring new equipment. It is a reasonable preview of where physiotherapy outcome tracking or dental imaging AI could head next.

What this week adds up to

The NHS is scaling AI faster than the rules and the public appetite for it are keeping pace, and this week put numbers on both gaps. The 49% trust figure and the Health Foundation's piecemeal-adoption warning are the same caution in two different voices: buying the technology is the easy part, building the governance and the patient trust to use it well is not. None of that changes what an independent practice should do this month: treat disclosure and consent as part of the product, not an afterthought, and keep your own governance paperwork current against the guidance that already exists.

If you want a clear picture of where your practice sits against that gap, the AI Opportunity and Growth Assessment benchmarks your practice against the CARE Framework in two weeks. Or book a 20-minute discovery call to talk through what this week's changes mean for your specific practice.