CASRAI Dictionary

Author: MCP Service

  • China AI Regulation for Research Collaboration

    China’s AI regulation centres on the Interim Measures for the Management of Generative Artificial Intelligence Services (effective 15 August 2023), which require AI-service providers to disclose AI use and forbid listing generative AI as a co-author. For Western universities collaborating with Chinese institutions, the rules affect authorship credit, cross-border data transfer, and how AI tools may be used in co-supervised research.

    China’s Interim Measures for the Management of Generative Artificial Intelligence Services is the country’s first binding, AI-specific regulation, jointly issued by the Cyberspace Administration of China (CAC) and six other ministries. It sits alongside the Cybersecurity Law, the Data Security Law and the Personal Information Protection Law (PIPL) as the legal backbone for how AI-enabled research involving Chinese partners must be conducted.

    This matters for research administrators well beyond China’s borders. Joint-authorship agreements, data-sharing memoranda and co-supervision arrangements with Chinese universities now have to reconcile Chinese disclosure and labelling duties with the authorship norms already in force under COPE, ICMJE and journal policy in the EU, UK and US.

    What China’s Interim Measures for Generative AI actually require

    The Interim Measures require providers of public-facing generative AI services to register with the CAC, prevent outputs that undermine state security or social stability, and take measures against algorithmic bias. Internal research and development that is not offered as a public-facing service is treated more lightly, but outputs intended for publication or public dissemination fall squarely within scope.

    Two further instruments extend the regime. The Measures for Labeling AI-Generated and Synthesized Content, paired with the national standard GB 45438-2025, took effect in September 2025 and require visible or embedded labels on AI-generated text, images and audio distributed in China. The Ministry of Science and Technology’s guidelines on responsible research conduct, issued in December 2023, apply specifically to academic work: they prohibit using generative AI to draft funding applications and require researchers to disclose any generative AI use in their methodology.

    China has not enacted a single, comprehensive AI statute. A draft Artificial Intelligence Law has appeared on the National People’s Congress Standing Committee’s legislative agenda since 2023, but no official draft had been released as of December 2025, and the enactment timeline remains unclear.

    How China’s framework compares with the EU, UK and US

    None of the four major jurisdictions regulates AI in research collaboration through a single dedicated instrument. Each layers AI-specific rules on top of existing data-protection, cybersecurity and research-integrity frameworks, but the point at which those rules bind differs sharply.

    Jurisdiction Core AI instrument Status (as of mid-2026) Authorship / disclosure rule for research
    China Interim Measures for Generative AI Services (2023) plus labelling rules (2025) In force; comprehensive AI Law still in draft Ministry of Science and Technology guidelines bar listing AI as a co-author; AI use must be disclosed
    European Union AI Act, Regulation (EU) 2024/1689 General-purpose AI obligations apply from August 2025; most other obligations from August 2026 No AI-authorship bar in the Act itself; publishers apply COPE and ICMJE norms
    United Kingdom No dedicated AI statute; pro-innovation, regulator-led approach Existing regulators (ICO and sector bodies) apply cross-cutting principles COPE- and ICMJE-aligned: AI cannot be listed as author; disclosure expected in methods sections
    United States No comprehensive federal law; state statutes (e.g. the Colorado AI Act) and the voluntary NIST AI Risk Management Framework Patchwork of state laws; federal approach still executive-order-driven NIH bars AI from being listed as an author or used by peer reviewers to evaluate applications; journals follow ICMJE/COPE

    The practical convergence is striking: China, the EU, the UK and the US all reach the same conclusion on authorship — a generative AI system cannot satisfy the accountability that authorship implies — even though none of them arrives there through identical legislation.

    What this means for joint authorship and contributor disclosure

    China’s Ministry of Science and Technology guidelines and the international consensus reflected in ICMJE recommendations and COPE position statements agree on one point: generative AI tools cannot be listed as authors or contributors, because they cannot take responsibility for the accuracy and integrity of the work. This aligns with the accountability criterion embedded in the CRediT contributor role taxonomy, which CASRAI originated in 2014 and which is now stewarded by NISO as ANSI/NISO Z39.104-2022.

    For joint publications with Chinese co-authors, this means AI-assistance disclosure statements now need to satisfy two regimes at once: China’s requirement to label AI-generated content and disclose AI use in the methodology, and the contributor-role documentation expected by journals following CRediT or ICMJE authorship criteria. A single disclosure paragraph, drafted to meet the stricter of the two standards, is usually sufficient — but it should name the specific generative AI tool, its role, and confirm that no tool is credited as an author or contributor.

    • Confirm which named human contributors meet Chinese and Western authorship criteria before drafting the manuscript.
    • Record AI-tool use (what, where, why) in a disclosure statement that satisfies both the Chinese labelling requirement and journal policy.
    • Never list a generative AI system as an author, co-author or contributor under any of the four frameworks compared above.

    Data-sharing and cross-border transfer requirements

    Research data moving out of China is governed by the Data Security Law and the Personal Information Protection Law, not by the Interim Measures themselves. Transfers of “important data” or bulk personal information generally require a CAC security assessment, a process legal trackers monitoring Chinese compliance report can take several months to clear. Projects that involve Chinese human genetic resources — common in biomedical and health-informatics collaborations — additionally require prior approval from the Ministry of Science and Technology before data can be shared internationally.

    Co-supervised doctoral projects that route data through a public-facing generative AI service add a further layer: the service falls within the Interim Measures’ registration and labelling scope, even where the underlying collaboration is privately arranged between two universities.

    Common questions on China’s AI regulation and research collaboration

    Does China have a comprehensive AI law?

    No. As of mid-2026, China has no single, comprehensive AI statute; regulation proceeds through targeted instruments — the Cybersecurity Law, the Data Security Law, the Personal Information Protection Law, and AI-specific measures such as the Interim Measures for Generative AI. A draft national Artificial Intelligence Law remains under review, with no confirmed enactment timeline.

    What is the Interim Measures for the Management of Generative AI Services?

    It is China’s first binding national regulation aimed specifically at generative AI, effective 15 August 2023. Issued jointly by the Cyberspace Administration of China and six ministries, it requires providers to register services, label AI-generated content, and prevent outputs that undermine state security or social stability.

    Can AI be listed as a co-author on Chinese-affiliated research?

    No. China’s Ministry of Science and Technology guidelines on responsible research conduct, issued in 2023, prohibit listing generative AI tools as co-authors and require disclosure of AI use in manuscripts and funding applications. This mirrors COPE and ICMJE guidance already applied by EU, UK and US publishers.

    Do foreign researchers need approval to share data with Chinese AI research partners?

    Often, yes. Under the Data Security Law and PIPL, transferring research data — especially human genetic or health data — outside China can require a Cyberspace Administration of China security assessment. Projects involving Chinese human genetic resources additionally need Ministry of Science and Technology approval before international sharing proceeds.

    Implications for research offices

    Research offices managing joint-authorship agreements, data-sharing memoranda or co-supervision arrangements with Chinese institutions need compliance processes that satisfy Chinese disclosure and security-review requirements without weakening the authorship and contributor-role standards already expected by Western journals and funders. Treating China’s rules as an additional layer on top of existing CRediT-based authorship practice, rather than a separate compliance track, keeps the paperwork proportionate.

    China’s regulatory posture is still moving: the Ministry of Science and Technology, the CAC and the State Council have all issued new instruments since mid-2025. Institutions with active China partnerships should treat authorship-disclosure and data-transfer procedures as living documents, reviewed annually against the current Chinese, EU, UK and US rules.

  • EU AI Act Compliance: University AI Checklist

    EU AI Act compliance obligations activate the moment a university’s AI system moves from a research prototype into real-world use. An admissions screening tool, a plagiarism detector, or a student-facing chatbot that starts operating on live applicant or student data falls outside the Article 2(6) research exemption and must meet the Regulation’s governance, documentation and human-oversight requirements for high-risk systems.

    The EU AI Act — Regulation (EU) 2024/1689 — is the European Union’s binding, risk-based framework that classifies artificial intelligence systems by risk level and imposes proportionate obligations on providers and deployers, including universities that operate AI tools within the EU or whose outputs affect EU users.

    What the Article 2(6) Research Exemption Actually Covers

    Article 2(6) excludes AI systems and AI models “specifically developed and put into service for the sole purpose of scientific research and development” from the Regulation’s scope. The exemption is narrow by design: it protects genuine R&D activity, not any AI project that happens to originate in a university lab.

    Most institutional coverage of the AI Act stops here, treating the research exemption as a blanket shield for higher education. It is not. The exemption tracks purpose, not origin: a model stays exempt only while its sole function is research — the instant it is repurposed to inform an operational decision, the exemption lapses for that use.

    This matters because universities routinely graduate tools from prototype to production: a thesis project becomes an admissions triage assistant, a plagiarism-detection experiment becomes the software every faculty uses to screen coursework. Each transition is a legal event under the AI Act, not just a technical rollout.

    Which University AI Systems Lose Exemption on Deployment

    Annex III of the AI Act designates four categories of education-sector AI as high-risk once deployed operationally: systems used to determine admission or assignment to an institution, to evaluate learning outcomes, to assess the appropriate level of education an individual should receive, and to monitor or detect prohibited student behaviour during tests — the Annex III wording that squarely covers exam-integrity and plagiarism-detection tools.

    A separate, already-enforceable rule applies to emotion-detection features sometimes bundled into exam-proctoring software: Article 5(1)(f) has banned emotion-recognition systems in educational institutions since 2 February 2025, with narrow exceptions for medical or safety purposes. A proctoring tool that infers stress or attentiveness from webcam data is not merely high-risk — it may be prohibited outright.

    Student-facing chatbots sit differently on the risk scale. A general enquiries chatbot typically falls under the lighter Article 50 transparency regime — it must disclose that users are interacting with AI — unless its outputs feed directly into an Annex III decision such as admissions ranking, in which case the high-risk obligations apply to that decision pathway.

    Deployment stage Example AI Act status University’s primary duty
    Lab prototype Model trained on institutional data, never used operationally Exempt — Article 2(6) Monitor for change of purpose
    Pilot with real users Admissions-triage assistant tested on live applicant files Conditionally exempt via regulatory sandbox Informed consent; sandbox registration (Article 57)
    Live admissions tool AI ranks or screens applicants operationally High-risk — Annex III(3)(a) Full Articles 9–15 obligations; FRIA (Article 27)
    Live exam-integrity monitor AI flags prohibited behaviour during tests High-risk — Annex III(3)(d) As above, plus an Article 5(1)(f) emotion-recognition check
    Public-facing chatbot Answers prospective-student enquiries Limited risk — Article 50 AI-interaction disclosure only

    The Governance and Documentation Checklist

    Once a system loses exemption, the deployer obligations that apply are the same ones any commercial organisation faces — but universities carry one duty that most private-sector guidance omits. Under Article 27, deployers that are bodies governed by public law must complete a Fundamental Rights Impact Assessment before putting a high-risk system into use. Most EU public universities meet that definition, which makes the FRIA a default step, not an optional extra.

    1. Inventory and classify every AI system reaching operational use, including vendor and embedded tools — not only in-house builds.
    2. Re-test Article 2(6) applicability at every go-live decision; log the classification rationale.
    3. Complete a Fundamental Rights Impact Assessment (Article 27) before deployment, particularly where the institution is a public-law body.
    4. Screen for Article 5 prohibited practices, including emotion recognition in educational settings.
    5. Establish human oversight checkpoints under Article 14: named staff, defined intervention points, escalation routes.
    6. Centralise technical documentation, instructions for use and event logging under Articles 11–13.
    7. Verify the vendor’s conformity assessment where a third-party tool is used — compliance cannot be outsourced to the supplier.
    8. Register the system in the EU high-risk database (Article 71) once the applicable Annex III deadline is reached.

    The compliance timeline has moved since most explainer content was written. Article 5 prohibitions and AI-literacy obligations have applied since 2 February 2025. General-purpose AI model obligations under Articles 51–55 have applied since 2 August 2025. Article 50 transparency duties take effect on 2 August 2026. Following the AI Act Omnibus political agreement of 7 May 2026, the Annex III high-risk deadline for use-based systems — including the education-sector list above — has been deferred to 2 December 2027, pending formal adoption and publication in the Official Journal.

    The deferral changes the runway, not the workload. Institutions that wait for the 2027 deadline to start classification and documentation work will find the FRIA and human-oversight design take longer to build than the calendar suggests.

    Compliance-Checker Tools and Regulatory Sandboxes

    The European Commission operates an official EU AI Act Compliance Checker through its AI Act Service Desk, which helps providers and deployers work out which obligations apply to a given system. It is a useful first-pass triage tool, but it does not substitute for a documented FRIA — it tells an institution which article applies, not how to evidence compliance with it.

    For institutions building a repeatable governance structure rather than a one-off assessment, ISO/IEC 42001 — the international standard for AI management systems — maps closely to the AI Act’s risk-management, data-governance and documentation articles, and offers a certifiable framework that research offices can run alongside existing research-integrity governance.

    Universities piloting a system before full operational rollout have a formal route available: Article 57 requires each Member State to establish at least one AI regulatory sandbox, giving providers — including public research institutions — a supervised environment to test systems with real users under national-authority oversight before the full high-risk regime applies.

    This governance shift sits alongside a broader move across research administration, where institutions are building the same kind of structured accountability for AI tools that they have long applied to research-integrity and data-management obligations.

    Common Questions on EU AI Act Compliance for Universities

    Does the EU AI Act research exemption cover university AI tools after deployment?

    No. The Article 2(6) exemption applies only while a system is developed and used solely for scientific research. Once a university deploys the same tool operationally — for admissions, plagiarism detection or another administrative decision — the exemption ends and high-risk or transparency obligations apply.

    Which university AI systems count as high-risk under the EU AI Act?

    Annex III lists four education categories: systems deciding admission or assignment, evaluating learning outcomes, assessing appropriate education level, and monitoring prohibited behaviour during tests. Admissions-screening tools and exam-integrity or plagiarism-detection systems fall squarely within this list once operational.

    What is a Fundamental Rights Impact Assessment and does it apply to universities?

    A Fundamental Rights Impact Assessment (Article 27) evaluates a high-risk AI system’s effect on individuals’ rights before deployment. It is mandatory for deployers that are bodies governed by public law — a category that covers most public universities in the EU deploying Annex III systems.

    When do EU AI Act high-risk obligations for education systems take effect?

    Following the Omnibus political agreement of 7 May 2026, Annex III high-risk obligations — including the education-sector list — are deferred to 2 December 2027, pending formal adoption. Article 5 prohibitions and GPAI obligations are already enforceable now.

    For research administrators, the practical implication is sequencing: build the AI inventory, classification log and human-oversight design now, while the Annex III deadline still allows time for a proper Fundamental Rights Impact Assessment rather than a rushed one. Waiting for the deadline to arrive before starting is the most common way institutions turn a manageable governance project into a last-minute compliance emergency.

  • Sovereign AI Fund: The University Research Route

    The UK’s Sovereign AI Fund is a £500 million state-backed venture capital vehicle, launched by the Department for Science, Innovation and Technology (DSIT) in April 2026, that makes equity investments and compute grants in British AI startups — it is not a university research grant scheme. University research groups instead access AI compute through the separate AI Research Resource (AIRR) open-access calls and funding through UK Research and Innovation (UKRI), routes this article sets out in detail.

    The sovereign ai fund operates like a professional venture capital firm with the balance sheet of the state behind it. Understanding where its remit stops — and where academic infrastructure routes begin — matters for any institution tracking funder and national-infrastructure policy in 2026.

    Contents

    What is the UK’s Sovereign AI Fund?

    The Sovereign AI Fund is a £500 million venture capital fund established by the UK government in April 2026 to invest directly in early-stage and growth-stage British AI companies. It sits within DSIT’s Sovereign AI Unit and was announced by Technology Secretary Liz Kendall and Chancellor Rachel Reeves as part of the government’s wider “AI maker, not AI taker” strategy first set out in the AI Opportunities Action Plan of January 2025.

    Equity cheques typically run from £1 million to £10 million, with the Fund’s own published materials citing up to £20 million for later-stage follow-on rounds. Portfolio companies also receive fully funded access to UK supercomputers — up to 1 million GPU hours per startup — plus fast-tracked visa decisions and help navigating procurement and regulation. The Fund is chaired by James Wise of Balderton Capital, with Suzanne Ashman appointed Managing Partner of its investment committee in May 2026.

    How do the Fund’s capital and compute tracks differ?

    The Sovereign AI Fund runs two distinct tracks, and conflating them is the most common misreading of the programme. The first is direct equity investment; the second is compute-only access to the AI Research Resource (AIRR) supercomputer network, awarded competitively without an immediate equity stake.

    By May 2026, three companies had received direct equity backing: Callosum (an AI infrastructure orchestration startup founded by Cambridge PhDs, and the Fund’s first investment), Ineffable Intelligence (founded by David Silver, former Head of Reinforcement Learning at Google DeepMind), and Isomorphic Labs (the drug-discovery company founded by Demis Hassabis). A further six companies — Cosine, Doubleword, Odyssey, Prima Mente, Twig Bio and Cursive — received AIRR compute allocations only, with the Fund holding a right of first refusal on future equity investment in several of them.

    • Equity track: capital plus compute plus visas, in exchange for a stake in the company.
    • Compute-only track: AIRR GPU allocation via a competitive open call, assessed on strategic relevance, technical quality and material compute need, with no immediate equity taken.
    • Strategic Assets Grants Programme: a separate £282 million pot funding shared datasets and infrastructure “critical inputs” for the wider AI ecosystem.

    Can university research groups access the Sovereign AI Fund directly?

    No. The Sovereign AI Fund is structured as a commercial venture-investment vehicle for UK-registered companies, not a research council grant scheme, and university departments are not eligible applicants in their own right. Academic groups seeking large-scale AI compute or project funding should instead route through the AI Research Resource and UKRI — mechanisms built for exactly this purpose and administered separately from the Fund.

    The AIRR network — which includes Isambard-AI at the University of Bristol and Dawn at the University of Cambridge — runs its own “AI Open Access” calls for academic-led projects requiring substantial GPU capacity in priority areas such as materials science, medical research and engineering biology. Eligibility generally requires the project lead to hold a lecturer-level (or equivalent) post at an organisation eligible for UKRI funding. Direct funding for AI-related research, meanwhile, flows through UKRI and its constituent councils, including the Engineering and Physical Sciences Research Council (EPSRC); UKRI has stated a £1.6 billion AI investment commitment across 2026–2030, and new AI research labs led by Oxford and UCL are set to receive up to £60 million in government funding.

    The overlap is real but indirect. Several portfolio companies have active university collaborations — Prima Mente works with Oxford, Imperial and Edinburgh on biological foundation models, and Callosum’s founders are Cambridge PhDs — but these run through standard knowledge-transfer channels, not Fund eligibility itself.

    Sovereign AI Fund vs AIRR vs UKRI: how the three routes compare

    Research administrators fielding questions from principal investigators or technology-transfer offices need a quick way to route enquiries correctly. The table below sets out the three mechanisms side by side.

    Mechanism Who it is for What it provides Cost to recipient
    Sovereign AI Fund (equity track) UK-registered AI startups £1m–£20m capital, up to 1m GPU hours, fast-track visas Equity stake taken by the state
    Sovereign AI Fund (compute-only track) Selected AI startups (competitive call) AIRR GPU allocation, no immediate capital None initially; right of first refusal on future investment
    AIRR AI Open Access University-led research teams (lecturer-level PI+) GPU time on Isambard-AI, Dawn and other AIRR nodes None — competitive academic allocation
    UKRI / EPSRC grants Eligible UK research organisations Project and infrastructure funding None — grant funding, no equity

    What does this mean for research administrators and institutional leaders?

    Institutions should treat the Sovereign AI Fund and AIRR/UKRI as two parallel but interlocking systems rather than one policy. Grants offices and research administrators should not point commercial spin-outs toward UKRI grant calls, nor point academic groups toward the Fund’s equity application form — the eligibility gates and outcomes differ fundamentally.

    There is also a capacity-planning implication. AIRR nodes such as Isambard-AI and Dawn now serve both academic open-access calls and Sovereign AI Fund-badged startup allocations from the same national compute pool. As the Fund plans to allocate compute “worth tens of millions of pounds” to startups this year, institutions relying on AIRR for research-council-funded work should factor potential contention into project timelines.

    Spin-out pathways deserve attention too. Academic teams that build a proof of concept using AIRR or UKRI-funded compute may later seek Sovereign AI Fund equity once they incorporate as a company — a legitimate sequence, but one that requires institutions to manage IP and data-rights handover clearly between the academic and commercial phases.

    Common questions about the Sovereign AI Fund

    What is a sovereign AI fund?

    A sovereign AI fund is a state-backed investment vehicle that deploys public capital, compute and strategic support into domestic AI companies. In the UK, this is the £500 million Sovereign AI Fund, which operates like a venture capital firm but is run by DSIT’s Sovereign AI Unit rather than a private investor.

    What exactly is sovereign AI?

    Sovereign AI refers broadly to AI capability — models, chips, data and infrastructure — that is built, controlled and hosted within a nation’s own jurisdiction rather than rented from foreign providers. The UK’s use of the term ties directly to the AI Opportunities Action Plan’s “AI maker, not AI taker” framing, adopted to reduce dependence on overseas AI infrastructure.

    Is Sovereign AI free to use for universities?

    The Sovereign AI Fund itself is not “free” — its equity track exchanges capital and compute for a stake in the company. For universities, the relevant comparison is AIRR’s Open Access compute calls and UKRI grant funding, both of which award GPU time or research funding without taking equity or ownership.

    What’s next for sovereign AI compute access?

    The Fund has confirmed it will keep assessing applications on a rolling basis and was, at its first cohort announcement, in discussions with around 30 further firms over AIRR access. The signal to watch is whether DSIT and UKRI publish a shared capacity-planning framework for AIRR, since academic and Fund-backed commercial demand now draw on the same national compute pool. Institutions that map their AI research pipeline against all three routes now, rather than after a bottleneck emerges, will be better placed as the 2026–2030 funding period unfolds.

    Institutions building AI-adjacent research programmes should track how funder infrastructure policy intersects with broader research administration practice, since compute-access rules now shape project feasibility directly.

  • AI Opportunities Action Plan: Research, Year One

    The AI Opportunities Action Plan, published by the UK Department for Science, Innovation and Technology (DSIT) on 13 January 2025, has met 38 of its 50 actions one year on, according to the government’s own “One Year On” progress report published 29 January 2026. For university research, delivery is real but uneven: new supercomputing capacity has landed, while AI Growth Zones and the Sovereign AI Unit’s research-facing funding remain mostly in the “designated but not yet delivered” phase.

    The AI Opportunities Action Plan is a 50-recommendation UK government strategy, authored by entrepreneur Matt Clifford, that commits the state to expanding compute infrastructure, unlocking public data assets, developing AI talent and accelerating public- and private-sector AI adoption. The government accepted all 50 recommendations in its January 2025 response and pledged a Compute Strategy for Spring 2025.

    Contents

    What compute has been delivered for university research?

    Compute is the section of the Action Plan with the clearest research-facing delivery record. The government committed £2 billion to expand UK public compute capacity twentyfold by 2030, and the first tranche has already reached campus-hosted infrastructure rather than staying at the announcement stage.

    • Isambard-AI, the flagship AI Research Resource (AIRR) supercomputer, launched at the University of Bristol in July 2025.
    • The DAWN supercomputer at the University of Cambridge was confirmed in January 2026 to receive a sixfold capacity increase, targeted for completion by Spring 2026.
    • A new national supercomputer backed by £750 million will be hosted in Scotland, coupled to the International Data Facility at the Edinburgh Parallel Computing Centre so researchers can run models against large datasets in a secure environment.
    • Up to £250 million has been earmarked specifically to scale cloud capacity within the AI Research Resource, the free-at-point-of-use compute pool for UK researchers, businesses and start-ups.

    This is the plan’s strongest evidence base: named machines, named universities and confirmed dates, rather than funding envelopes still awaiting allocation.

    Are AI Growth Zones and the Sovereign AI Fund reaching universities?

    Two of the plan’s highest-profile mechanisms — AI Growth Zones and the Sovereign AI Unit — show a wider gap between announcement and research-facing delivery than the compute programme does.

    Five AI Growth Zones have been designated across Great Britain, including two in Wales and one in Scotland, which the government reports have generated £28.2 billion in investment and more than 15,000 jobs, alongside £5 million of targeted local funding per zone. A new AI Growth Zone Delivery Unit has been created to broker power, planning and offtake agreements. But the government’s own document frames the coming year’s priority as “bringing AI Growth Zones from designation to delivery” — an explicit admission that build-out, not designation, is the unfinished task, and universities inside these zones are not yet reporting operational access to zone-linked infrastructure.

    The Sovereign AI Unit, backed by up to £500 million, has made a small number of research-adjacent commitments in its first year: it allocated sovereign compute to the University of Cambridge’s MACE materials-discovery foundation model, and provided £8 million in seed funding to the OpenBind consortium’s structural dataset for AI-driven drug discovery. The unit’s main investment phase — chaired by James Wise of Balderton Capital — does not launch until April 2026, meaning the bulk of its £500 million has not yet been deployed to UK AI companies or research spin-outs.

    Mechanism Committed funding Research-facing status, January 2026
    AI Research Resource / Isambard-AI, DAWN £2bn (20x compute by 2030), £250m cloud capacity Delivered — operational at Bristol, Cambridge scaling by Spring 2026
    Scotland national supercomputer + EPCC data facility £750m Committed, under construction
    AI Growth Zones (5 designated) £28.2bn investment reported, £5m per zone Designated; delivery unit only just established
    Sovereign AI Unit Up to £500m Early pilot investments only; main phase from April 2026
    Health Data Research Service Up to £600m (government + Wellcome) Leadership appointed Jan 2026; not yet operational

    What hasn’t been delivered yet?

    Twelve of the plan’s 50 actions remain unmet at the one-year mark. For research administrators, the most consequential gaps are structural rather than financial:

    • The AI Growth Lab cross-economy regulatory sandbox — intended to let promising AI applications, including research tools, trial in real-world settings ahead of full regulation — is still at the call-for-evidence stage, not operational.
    • The Health Data Research Service, jointly backed by government and the Wellcome Trust with up to £600 million, appointed its CEO (Dr Melanie Ivarsson) and Chair (Baroness Nicola Blackwood) only in late 2025 and January 2026 respectively; the single secure access point to national health datasets it promises is not yet live for researchers.
    • National Data Library funding of over £100 million has produced guidance and an open call for data proposals, but not yet a working data-sharing infrastructure that institutions can plug into.

    These are the items where the difference between “committed” and “delivered” matters most for institutions planning multi-year research infrastructure roadmaps.

    Answer-first: common questions on the Action Plan

    What is the UK AI investment plan?

    The UK’s core AI investment framework is the AI Opportunities Action Plan, backed by roughly £2 billion for compute expansion, a £500 million Sovereign AI Unit, and further sector funding through the 2025 Industrial Strategy and Spending Review 2025 settlements for AISI and the National Data Library.

    How much is the UK government investing in AI?

    Across the Action Plan’s first year, headline commitments include £2 billion for 20x compute capacity by 2030, £750 million for a new Scotland-based national supercomputer, up to £500 million for the Sovereign AI Unit, and £240 million for the AI Security Institute, alongside £600 million jointly with Wellcome for health data infrastructure.

    What are AI Growth Zones and do universities benefit?

    AI Growth Zones are five government-designated regions with streamlined planning and energy access to accelerate data-centre build-out. Universities within or near these zones have not yet reported operational research access, as the government itself states delivery — not designation — is the unfinished 2026 priority.

    What is the UK Sovereign AI Fund?

    The Sovereign AI Unit is a government-backed fund of up to £500 million designed to invest in and support UK AI companies across critical parts of the AI value chain. Its main investment phase, chaired by James Wise of Balderton Capital, begins in April 2026, after a first year of limited pilot allocations.

    What this means for research administrators

    Institutions should treat the Action Plan’s compute strand as substantially delivered and plan around it: AIRR access, Isambard-AI and the Cambridge DAWN expansion are real, usable capacity for 2026 research bids. AI Growth Zone and Sovereign AI Unit funding, by contrast, should still be treated as pipeline rather than available resource — research offices tracking institutional eligibility for zone-linked infrastructure or sovereign-fund co-investment should expect further delivery milestones through 2026 rather than immediate access. The Health Data Research Service is worth monitoring closely by any institution with health-data-dependent research programmes, given the scale of the £600 million commitment relative to its current pre-operational status.

    Outlook: the next year of delivery

    With 38 of 50 actions met, the government has moved the Action Plan from strategy document to partially built infrastructure. The test for its second year is converting designation into delivery — turning AI Growth Zones into working data-centre capacity, and the Sovereign AI Unit’s £500 million into deployed investment — while bringing the Health Data Research Service and National Data Library from governance milestones to infrastructure researchers can actually use. For university research administration teams, that distinction between committed and delivered funding will determine what can realistically be built into 2026–27 grant and infrastructure planning.

  • NIST AI Risk Management Framework for Research Offices

    The NIST AI Risk Management Framework (AI RMF) is a voluntary, four-function framework — Govern, Map, Measure, Manage — published by NIST in January 2023 to structure AI risk identification and mitigation across the system lifecycle, and it is increasingly the reference model research offices use to build AI-use policies for grant compliance and research computing.

    In one sentence: the NIST AI RMF is a voluntary, technology-neutral process framework — not a certification standard — that organises AI risk management into four continuous functions applied across governance, context-mapping, measurement and mitigation.

    What is the NIST AI Risk Management Framework?

    The NIST AI RMF (formally NIST AI 100-1) was directed by Congress under the National Artificial Intelligence Initiative Act of 2020 (P.L. 116-283) and published by the National Institute of Standards and Technology on 26 January 2023. It gives organisations a structured, repeatable way to identify, assess and manage AI-related risk without prescribing specific tools or vendors.

    Unlike a certification scheme, the AI RMF is deliberately flexible. Organisations apply it through “profiles” — documented mappings of the Core functions to a specific system, unit or risk context — supported by companion NIST materials including the AI RMF Playbook, Roadmap and sector Crosswalks. For a university research office, that flexibility matters: the same four functions can govern an AI-assisted grant-writing tool, a research-computing cluster running a locally hosted model, and a vendor’s generative-AI research assistant, each with a different risk profile.

    What are the four core functions — Govern, Map, Measure, Manage?

    The AI RMF Core is organised into four functions that operate continuously rather than sequentially: Govern establishes accountability and policy; Map identifies context and potential harms; Measure tests and monitors systems against trustworthy-AI characteristics; and Manage prioritises and resources mitigation. Each function contains categories and subcategories that a research office adapts rather than adopts wholesale.

    Function Purpose Typical research-office artefact
    Govern Sets accountability, policy and approval authority for AI use Institutional AI-use policy; PI attestation clause in proposal sign-off
    Map Documents context, stakeholders and where AI touches sponsored work Inventory of AI tools used in grant writing, review, and data analysis
    Measure Tests systems for validity, bias, security and privacy Vendor security questionnaire; bias check on AI-assisted scoring tools
    Manage Prioritises, mitigates and documents residual risk Incident log for AI-related data exposure; annual policy review

    The Core does not mandate a fixed maturity level. Organisations document which subcategories they have deferred, and why, alongside compensating controls — a discipline that maps onto existing research-compliance practices such as data management plans.

    What does NIST AI 600-1 add for generative AI?

    NIST AI 600-1, the Generative Artificial Intelligence Profile, was published in July 2024 as a companion to the AI RMF specifically for generative and foundation models. It does not replace the four-function Core; it applies Govern, Map, Measure and Manage to risks that are distinctive to generative systems.

    The profile documents risk across twelve categories, including confabulation (hallucinated outputs presented as fact), data privacy, harmful bias and homogenisation, information integrity, information security, intellectual property, and value-chain and component integration risk from third-party foundation models. For a research office, several of these map directly onto everyday research-computing and grant-compliance exposure:

    • Confabulation in AI-assisted literature review or preliminary-data summaries submitted in a proposal narrative
    • Data privacy exposure when researchers paste sponsor-restricted or human-subjects data into a public generative-AI tool
    • Intellectual property risk when proprietary or pre-publication research content is used as a prompt input to a third-party model that retains data for training
    • Information security gaps in export-controlled or ITAR-restricted research computing environments running locally hosted generative models

    How should research offices map RMF functions to grant compliance and research computing?

    Applying the AI RMF in a research office starts with an honest inventory, not a policy document. Most institutions already run parallel compliance regimes — IRB, export control, data use agreements, conflict of interest — and the AI RMF’s four functions slot into that existing governance architecture rather than requiring a new one.

    RMF function Research-office action Compliance touchpoint
    Govern Define who approves AI use in proposal preparation, peer review, and award administration Grant-compliance office; research integrity policy
    Map Inventory AI tools touching sponsor data, human-subjects data, or export-controlled research IRB, data use agreements, export-control review
    Measure Evaluate vendor AI tools for data retention, training-data use, and bias before procurement Procurement security review; research-computing vendor assessment
    Manage Maintain an incident-response path for AI-related data exposure or integrity failures Research integrity office; sponsor notification obligations

    Funders are beginning to require disclosure of AI use in proposal preparation and review; UKRI and the US National Institutes of Health have each issued guidance addressing generative-AI use in grant applications and peer review. A documented AI RMF-aligned policy gives a research office a defensible, auditable answer when a sponsor, an IRB, or an internal audit asks how AI risk is managed.

    How does the NIST AI RMF compare to ISO 42001 and the EU AI Act?

    The NIST AI RMF, ISO/IEC 42001, and the EU AI Act address the same problem — AI risk — through three different mechanisms, and international research offices often need to satisfy more than one at once.

    • NIST AI RMF: voluntary US guidance, published January 2023, no certification mechanism, technology-neutral
    • ISO/IEC 42001:2023: an internationally certifiable AI management system standard, published December 2023, auditable by an accredited body
    • EU AI Act (Regulation (EU) 2024/1689): binding law, entered into force August 2024, with risk-tiered obligations phasing in through August 2027 for high-risk systems

    Institutions with Horizon Europe funding, EU partners, or EU-based subsidiaries generally need to track the EU AI Act’s binding obligations separately from a voluntary AI RMF programme; the AI RMF’s four functions nonetheless provide a practical operational baseline that can be extended toward either ISO 42001 certification or EU AI Act compliance evidence without rebuilding the governance structure from scratch.

    Answer-first questions on the NIST AI RMF

    What are the seven steps of the NIST Risk Management Framework?

    The seven steps — Prepare, Categorize, Select, Implement, Assess, Authorize, Monitor — belong to NIST Special Publication 800-37, the general-purpose cybersecurity Risk Management Framework, not the AI RMF. The NIST AI Risk Management Framework uses a separate four-function structure (Govern, Map, Measure, Manage) with no authorisation-cycle requirement. Research offices should not conflate the two documents.

    What is the difference between ISO 42001 and the NIST AI RMF?

    ISO/IEC 42001:2023 is a certifiable AI management system standard that an accredited body can audit, published December 2023. The NIST AI RMF is voluntary US guidance with no certification mechanism. Many research offices use the AI RMF’s four functions to build the internal controls that ISO 42001 certification later formalises against an external auditor.

    What are the four types of AI risk?

    NIST’s AI RMF and its Generative AI Profile group AI risk broadly into performance risk (validity, reliability), societal risk (harmful bias, fairness), security risk (adversarial manipulation, data leakage), and third-party or value-chain risk from vendor models and training data. Research offices typically encounter all four simultaneously when adopting AI-assisted research tools.

    What are 5 risks of AI?

    For research administration specifically, the highest-priority risks are data privacy breaches in sponsor-data pipelines, confabulation in AI-assisted literature synthesis, intellectual property exposure through third-party model training on prompts, harmful bias in automated review or scoring tools, and information security gaps in procured generative-AI platforms.

    Implications for research administration

    The AI RMF’s voluntary status will not last as a governance shortcut. Grant-making agencies and international funders are moving toward AI-use disclosure requirements in proposal and reporting workflows, and institutions without a documented, RMF-aligned policy will increasingly answer ad hoc rather than from a defensible framework.

    Research offices already manage layered compliance regimes across research administration functions — export control, human-subjects protection, conflict of interest — and the AI RMF’s four functions sit inside that structure rather than replacing it. Starting with Govern (assign accountability) and Map (inventory AI touchpoints in sponsored work) gives most offices a defensible position within one administrative cycle, ahead of any future mandatory requirement.

  • ISO 42001 Certification for Research Offices

    ISO 42001 (ISO/IEC 42001:2023) is the first international standard specifying requirements for an Artificial Intelligence Management System (AIMS) — the governance framework an organisation puts around how it designs, procures, deploys and monitors AI. For a research institution, certification means running a documented, audited system of AI policies, risk assessments and human-oversight controls, not simply buying compliant software. It follows the same Plan-Do-Check-Act structure as ISO 27001, but its controls are built around AI-specific harms: algorithmic bias, opacity, data quality and misuse.

    ISO/IEC 42001:2023 is defined by the International Organization for Standardization as a management-system standard for establishing, implementing, maintaining and continually improving an AI management system within an organisation of any size or sector.

    Contents

    What is ISO/IEC 42001 and what does an AIMS cover?

    An AI Management System is a structured set of policies, roles, risk processes and records that governs how an organisation develops, procures or uses AI across its lifecycle. ISO/IEC 42001:2023 sets requirements for that system in its main clauses (4–10), plus an AI-specific Annex A control set — AI policy, resourcing, AI system impact assessment, data for AI systems, and third-party AI relationships.

    Unlike a product certification, ISO 42001 does not certify a specific model as “safe”. It certifies that the organisation has a working management system for whichever AI systems fall inside its declared scope — a research-grants triage tool, an admissions-screening system, or a plagiarism-detection service, for example.

    What does ISO 42001 certification actually involve?

    Certification is run by an accredited, independent certification body — in the UK, accreditation is overseen by the United Kingdom Accreditation Service (UKAS). The organisation implements first; the certification body then verifies.

    • Scope and gap analysis: define which AI systems, departments and data flows the AIMS covers, then assess current practice against ISO 42001’s clauses and Annex A controls.
    • AI system impact assessment: a formal review of the potential effects of each in-scope AI system on individuals and groups — bias, fairness, transparency, data provenance and human oversight.
    • Risk treatment and controls: implement policies, technical controls and role assignments (an “AI owner” is typically named for each system) to treat identified risks.
    • Internal audit and management review: test the system internally before the external audit and correct nonconformities.
    • Stage 1 audit: the certification body reviews documentation and AIMS design for readiness.
    • Stage 2 audit: the certification body tests whether the AIMS is operating effectively in practice, not just on paper.

    Once granted, certification is valid for three years, with annual surveillance audits to confirm the AIMS is still being maintained. This mirrors the certification cycle used for ISO 27001 and ISO 9001, since all three share the same Annex SL high-level structure.

    How does ISO 42001 differ from ISO 27001?

    ISO 42001 governs the management of AI systems; ISO 27001 governs the management of information security. They share the same clause numbering and audit mechanics, so organisations already certified to ISO 27001 typically find AIMS implementation faster — but the two standards are not interchangeable and neither certifies the other.

    Feature ISO/IEC 42001:2023 ISO/IEC 27001:2022
    Primary focus Governance of AI systems across their lifecycle Confidentiality, integrity and availability of information assets
    Distinctive controls AI impact assessment, data quality for AI, AI system life cycle, third-party AI relationships Access control, cryptography, physical security, supplier security
    Typical risk concerns Bias, opacity, misuse, unintended AI behaviour Breach, unauthorised access, data loss
    Structure Annex SL clauses 4–10 + Annex A Annex SL clauses 4–10 + Annex A
    Certification cycle 3 years, annual surveillance audits 3 years, annual surveillance audits

    In practice, most institutions treat ISO 42001 as an addition to an existing information-security baseline rather than a replacement for it — an AI management system without underlying information-security controls leaves the data feeding those AI systems unprotected.

    Does ISO 42001 satisfy EU AI Act conformity assessment?

    ISO 42001 certification does not, by itself, satisfy EU AI Act conformity assessment obligations for high-risk AI systems. Regulation (EU) 2024/1689 (the AI Act) entered into force on 1 August 2024, with obligations for high-risk systems applying progressively from 2 August 2026. The Act’s presumption-of-conformity mechanism (Article 40) attaches to harmonised European standards, which are being drafted separately by CEN-CENELEC Joint Technical Committee 21 — ISO 42001, an international rather than harmonised European standard, is not automatically one of them.

    This matters directly for universities. Annex III of the AI Act lists AI systems used to determine access or admission to education, or to evaluate learning outcomes, as high-risk by default. A university deploying an AI-assisted admissions or grant-triage tool is a “deployer” under the Act regardless of ISO 42001 status, carrying deployer obligations — human oversight, logging, incident reporting — regardless.

    What ISO 42001 does provide is a documented, auditable governance framework that maps cleanly onto many AI Act requirements — risk management, data governance, human oversight, technical documentation — making a future conformity assessment faster to prepare for, even though it is not a substitute for one.

    Is it worth pursuing for a research institution?

    For a research office or university IT/AI-governance function, the case for ISO 42001 rests less on legal necessity and more on institutional risk management and funder or partner assurance. Certification demonstrates that AI used in grant review, research-integrity screening, or student-facing systems is governed by a documented, externally audited process rather than ad hoc practice.

    Costs mirror any ISO management-system certification: staff time for gap analysis and internal audit, the certification body’s audit fees, and ongoing annual surveillance costs. Institutions already holding ISO 27001 (or ISO 9001), with a research administration function already handling risk registers and compliance documentation, will find the incremental lift smaller than a first-time management-system project.

    The pragmatic sequencing: map which AI systems are actually in scope (research-tools procurement, admissions, integrity-checking), run a gap analysis against Annex A, then decide whether formal certification adds enough external assurance value to justify the audit cost — before, not instead of, tracking the EU AI Act’s phased high-risk obligations, which apply irrespective of certification status.

    Common questions about ISO 42001 certification

    What is ISO 42001 certification standard?

    It is third-party verification that an organisation’s AI management system meets the requirements of ISO/IEC 42001:2023 — covering AI policy, risk treatment, impact assessment and continual improvement — confirmed through a two-stage audit by an accredited certification body and maintained via annual surveillance audits.

    What is the difference between ISO 27001 and ISO 42001?

    ISO 27001 manages information security risk (confidentiality, integrity, availability of data); ISO 42001 manages AI-specific risk (bias, transparency, data quality, human oversight) across an AI system’s lifecycle. Both share the same clause structure, so many controls and much documentation can be reused between them.

    Is ISO 42001 certification worth it?

    It is worth it where an institution needs demonstrable, externally audited AI governance for funders, partners or regulators — particularly if it already holds ISO 27001. It is less clearly worth it as a standalone first management-system project, given the audit cost and the fact that certification alone does not satisfy EU AI Act conformity-assessment duties.

    Is ISO 27001 mandatory in the UK?

    No. ISO 27001 is voluntary in the UK; it is not a statutory requirement under UK GDPR or the Data Protection Act 2018, though it is widely used to evidence the “appropriate technical and organisational measures” those laws require. The same voluntary status applies to ISO 42001 — no UK or EU law currently mandates it.

    AI governance of this kind sits within the broader discipline of research administration, where risk, compliance and data-governance functions increasingly have to account for AI tools used across the grant and research lifecycle.

    What this means for research offices next

    Expect ISO 42001 adoption in the research sector to track two forces: institutional risk appetite around AI-assisted decision-making, and the EU AI Act’s phased high-risk obligations landing through August 2026 and August 2027. CEN-CENELEC’s harmonised standards work will eventually clarify how far ISO 42001 conformity can be credited toward AI Act presumption of conformity — research offices tracking AI governance now will be better placed when that mapping firms up.

  • AI Security Institute UK: What the Rebrand Means

    The UK AI Security Institute (AISI) is the government’s frontier-AI testing body, renamed in February 2025 from the AI Safety Institute to signal a sharper focus on cyber-harms, national security and misuse risk rather than broader ethical questions such as bias. For universities, the practical mandate — pre-deployment model access, evaluation infrastructure, and grant funding via the Alignment Project — has not shrunk, but proposals now compete more strongly when framed around security-relevant risk.

    The AI Security Institute is a directorate of the UK’s Department for Science, Innovation and Technology (DSIT) whose mission, in its own words, is “to equip governments with a scientific understanding of the risks posed by advanced AI.” It sits inside government but is designed, in AISI’s own framing, “like a startup in the government.”

    What is the AI Security Institute, and how did it start?

    AISI traces its origins to the Frontier AI Taskforce, launched with an initial £100 million budget in April 2023. It was formally established as the AI Safety Institute at the AI Safety Summit held at Bletchley Park in November 2023 — the world’s first major intergovernmental gathering on frontier-AI risk. The institute now operates on £66 million of funding per financial year, plus long-term resourcing commitments from DSIT.

    Its core activities are unchanged by the rename: testing leading AI systems before and after public release, informing UK and allied policymakers on emerging capabilities, and running an open-source evaluation platform called Inspect that lets companies, governments and academics run standardised safety tests. AISI holds pre-deployment access agreements with Anthropic, Google DeepMind and OpenAI, giving it — and by extension its research partners — visibility into frontier models before the public sees them.

    Why was the AI Safety Institute renamed the AI Security Institute?

    The rename took effect in February 2025, reported first by Infosecurity Magazine on 14 February that year. Observers, including Wikipedia’s contributor consensus on the institute’s own entry, read the change as signalling that AISI would step back from broader ethical territory — algorithmic bias, freedom of speech in AI systems — and concentrate on the most severe, security-relevant harms: cyberattacks, biological and chemical weapons uplift, and loss of control over autonomous systems.

    The shift echoed a parallel move in Washington. In June 2025, the US AI Safety Institute was renamed the Center for AI Standards and Innovation (CAISI), with then-Commerce Secretary Howard Lutnick stating that AI evaluation should not be used “under the guise” of restricting innovation. The UK’s own rename predates that, but both reflect a broader 2025 pivot among Western AI-safety bodies away from precautionary, existential-risk framing and toward concrete national-security and economic-competitiveness mandates.

    AISI’s published research areas now read as a security taxonomy rather than a general safety agenda: Cyber Misuse, Safeguards, Alignment, Control, Autonomy, Human Influence and Societal Resilience. Each maps to a specific threat model government departments can act on, rather than an open-ended ethics brief.

    How does AISI fit into the International Network of AI Safety Institutes?

    The International Network of AI Safety Institutes was agreed at the AI Seoul Summit in May 2024 and held its first formal meeting in November 2024. Its founding members are the UK, the United States, the European Union, Japan, France, Singapore, South Korea, Canada, Kenya and Australia (Australia’s own AI Safety Institute was announced in November 2025, after the network’s launch). Kenya remains the only African member.

    Membership matters for universities in a practical sense: the network’s joint testing exercises — including a July 2025 evaluation exercise on AI-agent risks such as sensitive-data leakage — set shared technical standards that AISI then applies domestically. A university research group that aligns its evaluation methodology with AISI’s is, by extension, aligning with a standard that a further nine jurisdictions recognise.

    International Network of AI Safety Institutes — selected member bodies
    Jurisdiction Institute Established
    United Kingdom AI Security Institute (AISI) Nov 2023; renamed Feb 2025
    United States Center for AI Standards and Innovation (CAISI) Nov 2023; renamed Jun 2025
    European Union EU AI Office May 2024
    France INESIA Jan 2025
    Japan J-AISI Feb 2024
    Singapore Digital Trust Centre (AISI-designated) Renamed May 2024
    Canada Canadian AI Safety Institute Nov 2024

    What does the rebrand mean for university model-access and red-teaming partnerships?

    For institutions pursuing model-access agreements or red-teaming collaborations, the security framing changes what gets funded, not whether funding exists. AISI mobilises more than £15 million in grants through the Alignment Project, open to university and non-profit researchers globally, and its priority-access arrangement covers over £1.5 billion of compute through the UK’s AI Research Resource and exascale supercomputing programme — a resource pool researchers can draw on for evaluation-relevant work.

    Three practical shifts follow from the rebrand:

    • Proposal framing: research questions pitched around cyber-misuse, safeguard robustness or loss-of-control scenarios now map more directly onto AISI’s stated research areas than proposals framed around general-purpose ethics or bias auditing.
    • Compute and model access: AISI’s pre-deployment agreements with frontier labs give it privileged visibility that university partners can sometimes access via joint evaluation projects — but access is gated by relevance to AISI’s security-risk taxonomy.
    • Policy context: the UK’s AI Opportunities Action Plan, published 13 January 2025, commits to expanding sovereign AI compute capacity at least 20-fold by 2030 and created a Sovereign AI Unit with up to £500 million in funding — infrastructure that sits alongside, not inside, AISI’s own compute allocation, but which shapes the wider funding climate university research offices are now navigating.

    Research administrators should note that AISI’s grant and access programmes are administered separately from Research England and UKRI mainstream funding lines, so due-diligence and reporting requirements differ from a standard research-council award.

    Answer-first Q&A

    Did the UK change the name of the AI Security Institute?

    Yes. The UK’s AI Safety Institute was renamed the AI Security Institute in February 2025. The institute itself did not change its legal status or parent department — it remains a directorate of DSIT — but its public mission language and research priorities shifted toward cyber-harms and national-security risk.

    What exactly does “AI security” mean in this context?

    In AISI’s usage, AI security covers risks where advanced models are misused for cyberattacks, biological or chemical weapons development, or where systems act autonomously beyond human oversight. It is narrower than the earlier “AI safety” framing, which also covered algorithmic bias and broader societal harms.

    Who leads the AI Security Institute?

    Adam Beaumont, formerly GCHQ’s Chief AI Officer, is Interim Director. Jade Leung, the Prime Minister’s AI Advisor and a former OpenAI governance lead, serves as Chief Technology Officer. Ian Hogarth chairs the institute, and its advisory board includes AI researcher Yoshua Bengio.

    Who funds the AI Security Institute?

    AISI is funded directly by the UK government through DSIT, at £66 million per financial year, with long-term resourcing commitments. It separately mobilises over £15 million in external grant funding through the Alignment Project for researchers, including those at universities, working outside government.

    Implications for research administrators

    The safety-to-security rebrand is best read as a narrowing of mandate language, not a withdrawal from academic engagement. Universities seeking model-access or red-teaming relationships with AISI should expect proposals to be evaluated more explicitly against its published risk taxonomy — cyber misuse, safeguards, alignment, control, autonomy, human influence and societal resilience — than against a general AI-ethics brief.

    Institutions should also track the International Network of AI Safety Institutes’ joint testing exercises as a source of emerging shared methodology, since AISI’s domestic evaluation standards are increasingly set in coordination with nine other jurisdictions rather than unilaterally. As the UK’s sovereign compute build-out under the AI Opportunities Action Plan proceeds toward its 2030 target, research offices with evaluation, red-teaming or alignment capacity are positioned to benefit from both AISI’s own grant lines and the wider national compute expansion.

    CASRAI tracks research-administration implications of national AI-governance bodies as part of its broader coverage of the standards landscape; see the CASRAI Dictionary for related terminology and the research administration hub for adjacent policy explainers.

  • Wellcome Trust Open Access Policy vs Plan S and REF Requirements

    The Wellcome Trust open access policy requires immediate, embargo-free deposit of Wellcome-funded research articles in Europe PMC under a CC BY licence, restricts article-processing-charge funding to fully open-access venues from January 2025, and layers a separate data-sharing mandate on top of its OA rules — diverging in mechanics from both Plan S’s route-based minimum and REF 2029’s embargo-tolerant, lower-bar licensing floor.

    Wellcome is a UK-based biomedical research charity and a founding funder of cOAlition S, the international funder consortium that created Plan S in 2018.

    What does Wellcome’s open access policy require in 2026?

    Wellcome’s policy, in force since 1 January 2021 and tightened twice since, applies to all original research articles arising in whole or part from its funding. Three mechanics define it. First, the article must be deposited in Europe PMC and made freely available on the official publication date, with no embargo permitted. Second, authors must retain enough rights to apply a CC BY licence to the Author Accepted Manuscript — a mechanism known as rights retention — with CC BY-ND granted only by exception. Third, from 1 January 2025 Wellcome funds article-processing charges only in fully open-access journals or platforms; transitional funding for hybrid “read and publish” agreements ended in December 2024.

    A 16 January 2024 update added a fourth route: where neither the Version of Record nor the Accepted Manuscript can be made compliant, a CC BY-licensed preprint posted to a Europe PMC-indexed server before final publication now satisfies the policy. Scholarly monographs and book chapters submitted after 1 January 2021 fall under a related but separate Wellcome monograph policy, which permits a maximum six-month embargo — a materially different rule from the zero-embargo standard applied to journal articles.

    How Wellcome aligns with — and adds to — Plan S

    Wellcome has been a cOAlition S founding member since 2018, and its journal-article rules track Plan S’s core requirements closely: immediate access, a CC BY default, and no embargo. Both frameworks recognise the same three compliance routes — publishing in a fully open-access venue, self-archiving via rights retention in a repository, or publishing through a transformative agreement — and both use the shared Journal Checker Tool to let authors verify a venue in advance.

    Wellcome goes beyond the Plan S baseline in enforcement and scope. Plan S sets principles each signatory funder operationalises independently; Wellcome adds funder-specific detail Plan S does not itself mandate — the 2024 preprint route, a ban on OA block-grant funds paying hybrid APCs, and named sanctions (loss of lead-applicant eligibility, suspended grant payments) for non-compliance. Plan S does not prescribe monograph rules; Wellcome does, via its separate six-month-embargo monograph policy.

    Where Wellcome diverges from REF 2029’s open access rules

    REF 2029 — the UK’s national research assessment exercise, run by Research England and the other UK funding bodies — is not a Plan S signatory framework, and its open access requirements are structurally looser than Wellcome’s. Under the REF 2029 policy for outputs published between 1 January 2026 and 31 December 2028, journal articles and conference proceedings must be deposited within three months of publication, but embargoes are still permitted: up to six months for Main Panels A and B, and up to twelve months for Main Panels C and D. That is a reduction from REF 2021’s 12- and 24-month allowances, but it is not the zero-embargo standard Wellcome and Plan S apply.

    REF 2029’s licensing floor is also lower. While CC BY is the funding bodies’ stated preference, a CC BY-NC-ND licence — Non-Commercial, No Derivatives — meets the minimum requirement, versus Wellcome’s CC BY default with only narrow CC BY-ND exceptions. REF 2029 additionally excludes monographs, book chapters and scholarly editions from its open access scope entirely, whereas Wellcome applies its own (separate) embargo rule to those output types. The table below summarises the divergence.

    Requirement Wellcome (2026) Plan S / cOAlition S REF 2029
    Embargo (journal articles) None None 6 months (Panels A/B); 12 months (Panels C/D)
    Default licence CC BY (CC BY-ND by exception) CC BY CC BY preferred; CC BY-NC-ND meets minimum
    APC funding scope Fully OA venues only (from Jan 2025) Route-dependent, funder-operationalised Not an APC-funding body
    Compliance route Europe PMC deposit, rights retention, or CC BY preprint Gold OA, rights retention, or transformative agreement Repository deposit (green route) within 3 months of publication
    Monographs/book chapters In scope; max 6-month embargo Not prescribed by Plan S itself Out of scope for REF 2029
    Data sharing mandate Separate DMSP requirement Not part of core Plan S text Not part of REF open access policy

    Data sharing and rights retention: Wellcome’s additional layer

    Neither Plan S nor REF 2029 mandates data sharing as a condition of open access compliance; Wellcome does, through a policy that operates alongside — not inside — its OA rules. Wellcome’s Data, Software and Materials Management and Sharing Policy, updated 1 August 2024, requires funded researchers to submit an outputs management plan and to maximise access to research data with as few restrictions as possible. For research relating to public health emergencies, the policy requires quality-assured interim and final data to be shared as rapidly and as widely as possible, ahead of formal publication.

    • A Data Management and Sharing Plan (DMSP) is typically required at the application or award stage, not deferred to end-of-grant reporting.
    • The rights-retention statement authors must insert into subscription and hybrid-journal submissions is a Wellcome-specific compliance artefact — it is not required in the same form under REF 2029’s repository-deposit route.
    • Non-compliance with either the open access or the data-sharing policy can trigger the same sanction: ineligibility to apply as lead applicant on future Wellcome grants.

    This is the funder-specific compliance gap institutions most often miss: a paper can satisfy REF 2029’s repository-deposit rule and still fail Wellcome’s audit if the underlying dataset was not made accessible under the separate data policy.

    Frequently asked questions

    Does Wellcome allow any embargo on open access articles?

    No. Wellcome’s open access policy requires immediate deposit in Europe PMC with no embargo for original research articles. This is stricter than REF 2029, which permits six- or twelve-month embargoes depending on the assessment panel, and applies only to journal articles and conference proceedings, not to monographs.

    Is Wellcome Trust a Plan S funder?

    Yes. Wellcome has been a founding member of cOAlition S since 2018 and its 2021 policy was designed to align with Plan S principles. However, Wellcome operationalises those principles through its own mechanics — including a 2024 preprint-compliance route and named non-compliance sanctions — that Plan S itself does not mandate.

    Do REF 2029 open access rules apply to monographs?

    No. REF 2029’s open access policy covers only journal articles and conference proceedings with an ISSN; monographs, book chapters and scholarly editions are excluded from the current cycle, though UK funding bodies have signalled monograph requirements from the following REF exercise.

    Will Wellcome pay for open access publication in a hybrid journal?

    Not from January 2025 onward. Wellcome’s OA block grant now funds article-processing charges only in fully open-access journals or platforms; the transitional funding for hybrid “read and publish” agreements ended in December 2024.

    Implications for institutions and researchers

    Research administration teams managing multi-funder portfolios cannot apply one embargo or licensing rule across Wellcome, Plan S-aligned funders and REF 2029 — the three frameworks set genuinely different floors. A paper compliant with REF 2029’s CC BY-NC-ND minimum via green deposit can still breach Wellcome’s zero-embargo, CC BY-default rule if Wellcome funding is also acknowledged. Institutions need compliance checklists that track funder-specific mechanics, not a generic “open access” requirement, and should route Wellcome-funded outputs through the Journal Checker Tool before submission rather than after acceptance.

    The direction of travel across all three frameworks is convergence on stricter terms: REF’s embargo ceilings have already fallen once, UK funding bodies have flagged monograph open access for the exercise after REF 2029, and Wellcome’s data-sharing layer signals that funders increasingly treat open access and open data as linked obligations, not separate ones. Compliance processes built around funder-specific detail, not the lowest common denominator, will hold up best as these policies keep tightening.

  • Research England QR Funding vs Project Grants

    Research England QR funding is the block grant that Research England distributes to English universities based on the quality and volume of their research, as assessed by the Research Excellence Framework (REF). Unlike a competitive project grant, it is not tied to a specific proposal: institutions receive it as an annual lump sum and decide internally how to spend it, which is why it underpins long-term research capacity rather than individual projects.

    Quality-related (QR) research funding is the UK’s main formula-based block grant for research, allocated by Research England to higher education providers (HEPs) in England as part of the four-nation “dual support” system. For 2025 to 2026, Research England distributed £1,987 million in total QR funding — the largest single component of its £2,731 million combined research, knowledge-exchange and capital budget, according to Research England’s own grant-allocations basis publication (reference RE-P-2025-04).

    What is QR funding, and why does it exist?

    QR funding exists to give universities unrestricted, recurrent income for research rather than money tied to a single project. Research England, the Department for the Economy Northern Ireland, Medr (Wales) and the Scottish Funding Council each operate an equivalent block grant, and all four bodies use REF outcomes to inform their formulas. This “strategic institutional” funding sits alongside — not instead of — competitive grants from UKRI’s seven Research Councils, forming the UK’s dual support system for research.

    Because QR allocations are anchored to a periodic exercise rather than annual bidding rounds, they change slowly. Research England has said it is “seeking to maintain stability” in QR investment while REF 2021 outcomes remain the reference point ahead of REF 2029.

    How does Research England calculate QR funding?

    The QR formula weights each institution’s REF-assessed research quality and volume, then applies subject cost weightings and a London weighting before converting the result into cash. Mainstream QR — the largest QR element — totalled £1,303 million for 2025-26, including a London weighting calculated at 12% of mainstream QR, per Research England’s technical guidance for QR and HEIF allocations 2025 to 2026.

    QR is not a single payment. Research England’s 2025-26 budget breaks it into five funding streams:

    QR funding element 2025-26 budget What it funds
    Mainstream QR (incl. London weighting) £1,303m Core research quality/volume, weighted by REF outcomes
    QR research degree programme (RDP) supervision fund £344m Postgraduate research student supervision
    QR charity support fund £219m Overheads on charity-funded research
    QR business research element £114m Overheads on business-funded research
    QR funding for National Research Libraries £7m Five designated national-importance research libraries

    Source: UKRI, Research England grant allocations basis 2025 to 2026 (RE-P-2025-04), Table C. The Research Excellence Framework itself is run jointly by the four UK funding bodies roughly every three to seven years and involves more than 1,000 expert assessors across 34 subject-based panels.

    QR funding vs competitive project grants: what is the difference?

    QR funding is allocated by formula to an institution; competitive grants are awarded by peer review to a named investigator’s proposal. QR arrives every year regardless of whether a particular project succeeds; a project grant ends when the funded work is complete. This distinction is the whole point of dual support — one stream buys stability, the other buys targeted innovation.

    Feature QR funding (Research England) Competitive project grants (e.g. UKRI Research Councils)
    Allocation basis Formula, driven by REF quality and volume Open competition, peer review of a specific proposal
    Recipient The institution (HEP) The named investigator/project team
    Duration Recurrent annual block grant Fixed project term
    Use of funds Institutional discretion Restricted to approved project costs
    Application required No — based on REF and other formula data Yes — competitive proposal submission

    Institutions typically use QR’s flexibility for costs competitive grants will not cover: bridging funding between grants, early-career research time, shared equipment, and preparing REF impact case studies. The Russell Group has described QR as playing “an essential and unique role in achieving breakthrough research.”

    What changed in the funding formula for 2025-26 and 2026-27?

    The core QR formula did not change for 2025-26: Research England confirmed “no changes to the funding methods or weightings for any other elements of QR funding” beyond one reversion. The QR research degree programme supervision fund returned to its usual calculation method in 2025-26, after a temporary adjustment had applied in 2024-25.

    • Strategic institutional research funding (SIRF) review — the Department for Science, Innovation and Technology (DSIT) has asked Research England to review the robustness and value of flexible formula-based research funding on an ongoing basis running to 2030.
    • Transparency pilot — from autumn 2025, Research England began systematically collecting evidence on how institutions use their QR allocation, a shift reported by Times Higher Education in September 2025 as universities being “asked to explain how they spend millions of pounds received in quality-related (QR) funding.”
    • Knowledge-exchange formula adjustment — a related Research England formula stream, the Higher Education Innovation Fund, introduced a £500,000 allocation cap for new entrants in 2025-26; from 2026-27, HEPs previously constrained by that cap receive their full calculated allocation without the annual increase modifier applied.

    None of these changes alter the headline QR total, but together they signal closer scrutiny of how block-grant funding is spent — a planning-relevant shift for institutions relying on QR discretion.

    What does this mean for institutional research capacity planning?

    Because mainstream QR is re-based on REF outcomes rather than annual performance, institutions can forecast it several years ahead — but that stability window narrows as REF 2029 approaches and unit-level results begin to shift. Research administrators planning multi-year investments (research-space commitments, technician posts, early-career fellowships funded from QR) should treat the current REF 2021-derived allocation as a plateau, not a permanent baseline.

    The transparency pilot adds a second planning consideration: institutions should expect to document QR spend against outcomes, not just receive and allocate it internally. Research administration teams coordinating REF impact case studies, research culture initiatives and postgraduate supervision funding are best placed to own this evidence trail before it becomes a formal reporting requirement.

    Common questions about QR funding

    What is QR funding?

    QR funding is Research England’s main block grant for research, allocated by formula rather than by application. It is calculated primarily from Research Excellence Framework (REF) quality and volume scores, and unlike a project grant, it is not tied to specific research aims — institutions decide how to use it.

    How much is QR funding worth?

    Total QR funding was £1,987 million for the 2025-26 academic year, of which £1,303 million was mainstream QR, £344 million funded postgraduate research supervision, and the remainder covered charity- and business-funded research overheads and national research libraries, per Research England’s published 2025-26 grant-allocations basis.

    How does QR funding work?

    Research England converts each institution’s REF-assessed research quality and volume, weighted by subject cost and location, into a cash allocation paid annually. There is no application process; allocations shift only when an institution’s underlying data — such as REF results, postgraduate numbers, or research income — changes relative to the sector.

    What is the Research England Policy Support Fund?

    The Policy Support Fund is a separate strand of Research England’s strategic institutional funding, budgeted at £29 million for 2025-26, that supports universities in developing policy-related impact case studies and engagement ahead of future REF exercises — distinct from, but administered alongside, core QR funding.

    Outlook: what to watch before REF 2029

    QR funding will keep functioning as the UK’s most predictable research income stream in the near term, but three trends will shape how much institutional autonomy it retains: the ongoing SIRF review through 2030, the new spend-transparency expectations, and the approach of REF 2029, which will eventually re-base every institution’s mainstream QR allocation. Institutions that build evidence of QR outcomes now — rather than waiting for reporting requirements to formalise — will be better positioned when the formula next resets.

    For research administration teams tracking how funder policy changes intersect with institutional compliance and reporting obligations, monitoring Research England’s annual grant-allocations publications alongside broader research administration developments remains the most reliable way to anticipate formula shifts before they land.

  • Athena Swan Charter Explained: UK Requirements

    The Athena Swan charter is a UK-originated equality accreditation framework, now run globally by Advance HE, that assesses how universities and research institutions advance gender equality. Institutions build a self-assessment case against ten charter principles and submit it for Bronze, Silver or Gold recognition, each award now valid for five years under the transformed framework introduced in 2021. This explainer sets out what the charter actually requires, tier by tier.

    The Athena Swan charter is an equality-charter accreditation scheme, established in 2005 by the UK’s Equality Challenge Unit (ECU) and now administered by Advance HE, that recognises institutional commitment to gender equality in higher education and research.

    What is the Athena Swan charter and who runs it?

    The Athena Swan Charter (Scientific Women’s Academic Network) began in 2005, established by the UK’s Equality Challenge Unit to encourage and recognise commitment to advancing women’s careers in science, technology, engineering, mathematics and medicine (STEMM). The first awards were conferred in 2006. The ECU merged into Advance HE in 2018, which now runs the charter as a global framework covering the UK, Ireland and affiliated schemes elsewhere.

    In May 2015 the charter expanded beyond STEMM to cover arts, humanities, social sciences, business and law (AHSSBL) departments, and extended eligibility to professional, support and technical staff as well as transgender staff and students. The first non-STEMM awards were announced in April 2016.

    What are the ten Athena Swan charter principles?

    Every signatory institution commits in writing to ten principles before submitting its first application. In summary, institutions commit to:

    • Recognising that academia cannot reach its full potential unless it benefits from the talents of all.
    • Advancing gender equality and addressing the loss of women across the career pipeline.
    • Addressing unequal gender representation across disciplines, including AHSSBL and STEMM specifically.
    • Tackling the gender pay gap.
    • Removing obstacles at major career transition points, including the move from PhD to a sustainable academic career.
    • Addressing the negative effects of short-term contracts on retention and progression, particularly for women.
    • Tackling discriminatory treatment experienced by trans people.
    • Securing active leadership commitment from senior levels of the organisation.
    • Making sustainable structural and cultural change rather than relying on individual-level initiatives alone.
    • Considering the intersection of gender with other identity factors wherever possible.

    These principles, confirmed by both Advance HE members and independently verified against the London School of Economics’ published Athena Swan documentation, have drawn criticism for not addressing collective bargaining or unconscious bias in “market rate” pay-setting — a gap noted in published academic critique of the scheme.

    How does the transformed charter change requirements?

    Advance HE published new transformed charter principles in November 2020, with institutions applying under the updated rules from 2021. Advance HE describes this as a shift “from prescription to autonomy and flexibility,” giving applicants more discretion over which priority areas to evidence rather than a fixed checklist.

    The transformed framework introduces a standardised departmental culture survey, clearer award-level criteria and a new assessor scoring rubric intended to improve consistency between panels. Advance HE states the streamlined process reduces the administrative burden on applicants by more than 50% compared with the legacy charter. Freedom of speech and academic freedom are now explicitly referenced within the charter’s scope, following an Advance HE consultation on the two Equality Charters it runs.

    What are the Bronze, Silver and Gold tiers, and what evidence do they need?

    Institutions and individual departments apply separately for one of three award levels. Each requires a Self-Assessment Team (SAT), quantitative and qualitative data analysis, and a SMART (specific, measurable, achievable, relevant, time-bound) action plan submitted for independent peer review.

    Award level Core requirement Typical evidence Validity
    Bronze Foundational self-assessment and commitment Staff/student data, SAT formation, baseline culture survey, initial action plan 5 years (transformed framework)
    Silver Embedded practice and demonstrated progress Evidence of Bronze action-plan delivery, sustained data trends, staff consultation 5 years (transformed framework)
    Gold Sector-leading, sustained impact Multi-year outcome data, evidence of influence beyond the institution, mature governance 5 years (transformed framework)

    Validity periods have changed over time: awards granted under the pre-2015 charter ran for three years; post-2015 awards ran for four years; the current transformed framework sets a uniform five-year validity before renewal or progression to the next tier, per guidance published by the University of Oxford’s Equality and Diversity Unit.

    Does the Athena Swan charter actually change outcomes?

    Evidence on impact is mixed, which most institutional Athena Swan pages omit. A 2020 retrospective cohort study found faster growth in female representation in managerial leadership among Athena Swan members. Separately, a 2020 BMJ study linked Athena Swan funding incentives at National Institute for Health Research Biomedical Research Centres to a rise in women in mid-level research leadership.

    Against that, empirical research from the University of Bath found no evidence that Athena Swan membership or award level affects the gender pay gap or the proportion of women in the top pay quartile. A 2025 review by the policy group Murray Blackburn Mackenzie similarly found a lack of evidence that the scheme has been effective at addressing sex inequalities in promotion or pay. Separate commentary reported by The Times in November 2021 raised academic-freedom concerns about the charter’s methodology, prompting the freedom-of-speech clarifications now built into the transformed framework.

    Internationally, Ireland runs its own 2021 Athena Swan Ireland charter, overseen by the Higher Education Authority; as of the April 2025 assessment round it had 148 award holders — 120 Bronze and 28 Silver. Related schemes adapting the methodology include Australia’s SAGE programme (2015), the US AAAS-supported SEA Change programme (2017) and Canada’s Dimensions programme (2018).

    Common questions about the Athena Swan charter

    Why is it called the Athena Swan charter?

    The name combines the Athena Project and the Scientific Women’s Academic Network (SWAN), two UK initiatives that merged in 2005 to promote women in science, technology, engineering, mathematics and medicine, before the charter later extended sector-wide to all disciplines and roles.

    How long does an Athena Swan award last?

    Under the current transformed framework, Bronze, Silver and Gold awards are valid for five years before an institution must renew or apply for the next level. Earlier charter rules gave four-year validity for awards made after 2015, and three years before that.

    Who runs the Athena Swan charter today?

    Advance HE, the UK higher-education sector body formed in 2018 from the merger of the Equality Challenge Unit, the Higher Education Academy and the Leadership Foundation, administers the charter. The Equality Challenge Unit originally established it in 2005.

    What are the Athena Swan charter award levels?

    Institutions and departments apply for Bronze (foundational self-assessment and commitment), Silver (embedded practice with demonstrated progress) or Gold (sector-leading, sustained impact), each requiring self-assessment evidence and a SMART action plan reviewed by an independent panel.

    What this means for research administrators and institutional leaders

    For research-administration teams, the transformed charter shifts the workload from a prescriptive evidence file toward an ongoing culture-survey and data-governance cycle, since Advance HE’s reduced-burden process assumes institutions already track staff and student equality data continuously rather than compiling it retrospectively. Self-Assessment Teams should treat renewal as a standing governance function, not a periodic project, given the five-year cycle and the panel’s expectation of demonstrated progress rather than a one-off snapshot.

    Institutional leaders should weigh the mixed effectiveness evidence honestly in governance reporting: citing award status as a proxy for measurable pay or promotion equity overstates what the published research currently supports.

    Outlook: the charter at 20 years

    Advance HE is marking 2025–26 as the charter’s 20th anniversary, using the milestone to review impact and gather sector feedback. With the transformed framework still bedding in since 2021, and independent reviews continuing to question its measurable effect on pay and promotion outcomes, institutions applying now should expect continued refinement of evidence standards rather than a static rulebook.

    Research administrators tracking equality, culture and governance frameworks alongside standards work such as CRediT and authorship policy can find related context in CASRAI’s research administration coverage.