Three AI Tools UK Doctors Lost in Six Weeks — And What Works in Their Place

In six weeks between late April and May 2026, three of the most credible clinical AI tools in the world became unavailable to UK NHS clinicians. ChatGPT for Clinicians launched with the UK and EEA excluded at the verification step. OpenEvidence withdrew from the UK and EU entirely. Heidi Evidence — UK-partnered with NICE, BMJ Group, MIMS, and HealthPathways — blocked NHS email addresses at enrolment.

If you are a UK NHS clinician, you may not have noticed all three. They happened quietly, in adjacent news cycles, each treated as a one-off rather than a pattern. The pattern is the point.

This post sets out what happened, why it happened, and — more usefully — what UK clinicians can actually use today.

The pattern, in dates

• 23 April 2026 — OpenAI launches ChatGPT for Clinicians. Free for verified US physicians, NPs, PAs, and pharmacists. UK and EEA explicitly excluded at the regional-verification step.
• 28 April 2026 — OpenEvidence withdraws from the UK and EU. No replacement announced. No UK handoff. The platform that 40 % of US physicians were reportedly using disappears from UK access overnight.
• May 2026 — Heidi Evidence (the clinical decision support arm of Heidi Health) confirms it is blocking UK NHS email domains at enrolment. Australian, US, and Singapore clinicians can sign up. UK NHS staff cannot.

We have written individual explainers for each: the OpenEvidence withdrawal, the ChatGPT for Clinicians exclusion, and the Heidi Evidence NHS-email block. This post pulls the thread.

Why this is happening

The three withdrawals look like coincidences. They are not. They share a single underlying cause: UK and EU regulatory uncertainty around clinical AI, at exactly the moment the products were ready to ship.

Each of the three tools generates AI-synthesised clinical recommendations from a clinical query. Under UK MHRA classification, that kind of output sits close to — and in some cases inside — Software as a Medical Device territory. Content partnerships with NICE or BMJ Group describe what a model is grounded in. They do not describe how a regulator classifies it.

The UK regulatory framework for AI in healthcare is in mid-flight. The Medical Devices (Amendment) Regulations consultation closes 19 June 2026. The International Reliance Framework is planned for Autumn 2026. The National Commission on AI in Healthcare report is due later in the year. The EU AI Act's high-risk classification for certain clinical AI systems applies in parallel.

For a US vendor about to launch a global clinical-AI product, the safest decision is to ship into markets where the regulatory pathway is clear (the US, certain Asian markets, Australia under TGA) and defer the UK and EU until the rules settle. None of these companies has cited the UK as a strategic concern. The UK has been deferred while the rules are being written.

The asymmetric effect on UK clinicians is what makes the pattern matter. Three credible options for clinical Q&A vanished in six weeks. Each disappearance was rational from the vendor's perspective. The cumulative effect on a UK NHS doctor's working week is real.

The three withdrawals, briefly

ChatGPT for Clinicians (23 April 2026)

OpenAI's clinician-facing product. Free for verified US healthcare professionals. Features include clinical search with citations, reusable workflow skills, deep medical research, documentation assistance, and CME credits. UK and EEA users navigating to the product receive a "not available in your region" message at the NPI-verification step.

OpenAI has not announced a UK launch date. Given the regulatory backdrop, none should be expected before the second half of 2026 at the earliest.

OpenEvidence (28 April 2026)

OpenEvidence raised $250 million in January 2026 at a $12 billion valuation. It offered AI-powered clinical reasoning with direct access to NEJM content, plus DeepConsult (multi-step clinical reasoning) and Coding Intelligence. In the US, the company claimed adoption by 40 % of practising physicians.

The UK and EU withdrawal cited AI regulatory uncertainty. No re-entry signal has been published. UK doctors who were trialling the platform have lost access; the few who built clinical workflows around it have had to reconfigure.

Heidi Evidence (May 2026)

Heidi Evidence is the most strategically interesting move in the trio. Free for individual clinicians. Citation-backed. Partnered with NICE, BMJ Group, MIMS, and HealthPathways — the four most credible UK clinical-content sources a tool in this space could pair with. The product is good. We have read the output carefully.

UK NHS clinicians cannot use it. The enrolment flow blocks email addresses on the major NHS domains (nhs.net, nhs.uk, NHS Wales, NHS Scotland, HSC Northern Ireland). Australian, US, and Singapore clinicians can sign up. The block is specific to Heidi Evidence; Heidi Scribe (the AI clinical scribe product, separately MHRA-cleared) remains available to UK clinicians, including NHS staff.

Heidi Health has not publicly explained the NHS block. The most likely explanation is the same regulatory caution that took OpenEvidence and ChatGPT for Clinicians out of UK access — but applied at the user-domain level rather than the geographic level.

What UK NHS clinicians can use today

There is a credible UK setup for clinicians who want both point-of-care Q&A and weekly literature current-awareness. It is not a workaround. It is a two-tool combination that does the same job the three withdrawn products were attempting, with the advantage that it is actually available.

For clinical Q&A at the point of care: iatroX

iatroX is the UK-built clinical-AI pull-tool that survived the same six-week period that took the other three out. It is MHRA-registered as a Class I medical device, carries UKCA marking, is UK-built, and retrieves answers grounded in NICE, CKS, and SmPC — the same sources a UK GP, consultant, or trainee would check by hand. Base tier is free for individual clinicians; paid specialty diplomas around £99/year. It includes calculators and Q-banks alongside the core retrieval product.

iatroX is not a perfect substitute for the model quality of ChatGPT for Clinicians or the corpus access of OpenEvidence. But it is the UK-regulatory-compliant tool that UK NHS clinicians can actually use today, and the regulatory ground it stands on is the ground the three withdrawn products could not.

We have no commercial relationship with iatroX. Mention is editorial. We point to iatroX because, in May 2026, it is the credible UK pull-tool option a UK NHS clinician can use.

For AI clinical documentation: Heidi Scribe

If you have been using Heidi Scribe (separate from Heidi Evidence), you keep it. Heidi Scribe is independently MHRA-cleared and is unaffected by the Heidi Evidence enrolment block. The two products share a brand and a parent company; they have different regulatory statuses and different UK availability.

For point-of-care reference: UpToDate (standard)

Standard UpToDate remains available to UK clinicians and is widely provided through NHS institutional access. UpToDate's new Expert AI tier — the natural-language Q&A layer — is currently restricted to the US and Canada for individual subscribers, with limited Enterprise Edition availability elsewhere. UK individuals cannot use Expert AI. The standard product is unaffected.

For staying current with the medical literature: a weekly digest

This is the half of the problem that pull-tools never solved, and the half that quietly matters most.

The Monday Clinical Brief is built for this role. Every Monday morning, you get every new peer-reviewed paper published in the past week from the five highest-impact journals in your specialty — summarised in plain language, with UK context (NICE, Royal College, NHS pathway) noted where it matters. Typically 10–20 papers per specialty per week, with significant variation by specialty and week. Thirty-one specialties. £20 per year. No advertising. No sponsored content. No regulatory uncertainty — we are a literature digest, not a medical device, and we do not claim to be one.

We have written separately about how iatroX and a weekly digest combine into a 30-minute-a-week workflow. The short version: a pull tool handles the questions you bring to it; a push tool tells you what to ask in the first place. Most UK clinicians benefit from one of each.

The two-tool workflow

A practical structure for UK clinicians who want both halves of the problem solved:

Monday morning — read the digest. Fifteen minutes. See what was published in your specialty last week. Flag two or three to read in full if the topic is relevant. Log the reading as CPD.

During the week — use iatroX for the questions that come up. A patient presents with something unfamiliar. A colleague asks a question you cannot answer from memory. A guideline you used to know has changed. iatroX gives you a UK-guideline-grounded answer in seconds.

Friday afternoon — five minutes to log and reflect. What did you read? What changed in how you think about it? One sentence on each. CPD evidence in five minutes. Use the MCB CPD Tracker if you want the workflow built around exactly this rhythm; FourteenFish if you already use it.

Total time: under 30 minutes most weeks, with the pull-tool time scaling to however many real clinical questions arise.

What happens next

The pull-tool exclusions will probably ease as the UK regulatory framework lands later in 2026. The MHRA consultation closes in June. The International Reliance Framework is planned for Autumn. The National Commission report is due in the second half of the year. A realistic best case is a clearer regulatory pathway by Q1 2027, with one or two of the currently-withdrawn tools returning to the UK once that pathway is visible.

In the meantime: the credible UK setup is iatroX for pull, a weekly digest for push, and either UpToDate (institutional) or BMJ Best Practice (NHS-bundled) for traditional reference. That covers the use cases the three withdrawn products were trying to occupy.

Our honest position

We run The Monday Clinical Brief. We are not neutral on the value of a weekly literature digest in this picture. But the recommendations in this post are not paid placements. We have no commercial relationship with iatroX. We have no commercial relationship with UpToDate or BMJ Best Practice. We mention each one because, in May 2026, they are the UK options that work — for the use cases the withdrawn products were trying to fill.

If the situation changes — if Heidi Evidence opens UK NHS access, if OpenEvidence returns to the UK, if ChatGPT for Clinicians launches a UK tier — we will update this post and say so explicitly. Until then, the six-week pattern is what UK doctors need to understand, and the two-tool setup above is the practical response.

Subscribe to The Monday Clinical Brief for the literature half of the problem. £20 a year. Four-week free trial. Thirty-one specialties.