Medical AI literacy is being standardised four times at once, and the differences are as instructive as the overlap. The UK's Medical Schools Council, working with HDR UK, has recommended preparing students in health informatics, AI and data science, spanning data governance and ethical, professional, legal and regulatory dimensions. The AAMC's competency-development work is shaping US expectations while adoption races ahead, 53% of US and Canadian schools in 2023 to 77% in 2024. Canada is building institutional infrastructure, including the June 2026 AFMC partnership with ScholarRx spanning all 19 faculties of medicine around modular English- and French-language open educational resources. And the AMC's Digital Capability Framework addresses the continuum of education and practice across Australia and New Zealand. Four frameworks, one generation of students, and this comparison maps what travels and what does not.
The shared core
Strip the branding and the four converge on six capabilities. Critical appraisal: reading AI outputs and AI evidence with calibrated scepticism, including knowing which outcome a study actually measured, /blog/ai-learning-outcome-ladder-medical-education. Bias and equity: understanding whose data trained the model and who its errors will find. Confidentiality: the never-paste rules and their reasoning. Human oversight and accountability: the clinician's responsibility surviving every tool. Communication: explaining AI involvement to patients and colleagues. Data literacy: enough statistical and informatics grounding to supervise rather than merely operate. A student strong on these six is prepared for any of the four systems, which is the reassuring half of the comparison; the national layers are where preparation must localise.
The national differences
UK: literacy lands inside NHS governance, NICE and national-guideline context, MHRA device categories and GMC accountability, so the UK-flavoured skill is knowing how a tool's regulatory status and a doctor's professional responsibility interlock, and UKMLA preparation increasingly rewards uncertainty-handling and patient-centred framing. US: the institutional and liability environment dominates, board structures, institutional policy variation, malpractice context, and the AAMC competency direction plus rapid curricular adoption means US students will meet AI as assessed content sooner than most. Canada: the distinctive layers are bilingual delivery, the population-health and Indigenous-health contexts a responsible curriculum must carry, and the emerging open-infrastructure ecosystem, faculty-facing authoring and shared resources alongside consumer tools, with student behaviour already ahead of policy, 78.9% using generative AI and 53% weekly in Ontario survey data. Australia and New Zealand: the AMC framework's continuum framing, rural and distributed training realities that strengthen AI's access case while sharpening its equity risks, and Australian privacy law as the placement-adjacent rulebook, the fuller treatment at /blog/australia-medical-students-ai-paradox-2026.
What this means for mobile students and for schools
For the growing population studying in one country and examining in another, the practical synthesis: build the shared core once, it transfers whole; localise three things deliberately, the guideline ecosystem, the regulatory-accountability structure and the examination's emphasis, because those are where a correct-elsewhere answer fails here, the jurisdiction problem in its educational form. For schools, the comparison suggests humility and theft in equal measure: none of the four frameworks is complete, each has a component the others under-weight, UK regulatory interlock, US assessment integration, Canadian equity infrastructure, Australasian continuum framing, and a curriculum committee could do worse than building from the union rather than the local default. The convergence is real, the core is stable, and the students it serves will practise across all four systems in one career anyway.
Frequently asked questions
Which country is furthest ahead?
Ahead on different axes: US adoption speed, UK regulatory clarity, Canadian infrastructure, Australasian framework breadth; the honest league table has four first places and four gaps.
Do these frameworks change what students should do this year?
They confirm it: master the shared core through use and appraisal, learn your jurisdiction's regulatory layer, and let unaided, blueprint-mapped performance remain the endpoint of any AI-supported study.
Where do platforms like iatroX sit relative to the frameworks?
As environments the literacy gets practised in, exam-mapped across UKMLA, USMLE, MCCQE and AMC tracks, while the frameworks themselves belong to schools and regulators; a platform that claims to teach the whole competency is overreaching by design.
Will the four frameworks converge into one?
The cores already have; the national layers are structural, regulation, language, geography, liability, and will persist, which is why mobile students should plan on one core plus deliberate relocalisation rather than waiting for a universal syllabus.
Which framework best serves an IMG moving between systems?
The shared core plus the destination's regulatory layer: an IMG heading to the UK, for instance, gains most from the MHRA-GMC interlock and NICE ecosystem, since the transferable six will already have been built wherever they trained.
Do postgraduate frameworks differ from these student-facing ones?
They extend rather than replace: the same six capabilities reappear in professional guidance with supervision and accountability weighted more heavily, which is exactly why building the core as a student pays compound interest.