As artificial intelligence becomes more deeply embedded in education, responsible adoption depends on more than what AI can do. It requires clear principles for how it should be used, safeguards for how data is handled and a commitment to keeping educators and institutions in control. Two new Avallain publications, developed as part of the Avallain Lab’s work, bring these considerations together, connecting the principles of trustworthy, human-centred AI with the technical measures needed to put them into practice.

From Principles to Protection: Building a Practical Framework for Responsible AI in Education

St. Gallen, September 2026 — As publishers, educational institutions and educators expand their use of artificial intelligence, questions around ethics, human oversight, privacy and security are becoming increasingly practical. It is no longer enough to ask whether an AI-enabled tool can support teaching, learning or content creation. Organisations also need to understand how that technology is designed, where responsibility sits, what happens to their data and how human judgement is preserved.

Two new Avallain publications address these questions from complementary perspectives: the Avallain Position Paper on AI Ethics and Security in Education and the Avallain AI Trust Model.

The Position Paper establishes Avallain’s view on the ethical, secure and practical use of AI in education, covering areas including human oversight, risk classification, transparency, bias, content safety, information security, data protection and the protection of learners. The AI Trust Model moves from these principles to technical implementation, outlining a multi-layer approach to protecting personal data across generative AI workflows.

Together, they provide a clearer picture of what responsible AI means not only as a principle, but as a responsibility spanning design, governance and operations.

Why Responsible AI Needs Both Principles and Practice

For organisations adopting AI in education, the challenge is not simply technological.

Publishers need to understand how to integrate AI into professional content workflows without compromising intellectual property, personal data or editorial responsibility. Institutions need confidence that AI-enabled systems can be introduced without weakening safeguarding, governance or educators’ professional authority. Teachers and other professional users need to understand what AI can help them accomplish and where its limitations require human review.

The wider educational sector faces the same challenge at scale. Generative AI can support content creation, lesson planning, feedback, assessment preparation and many other professional activities, but its outputs can also contain errors, bias, outdated information or content that does not reflect a particular pedagogical or curricular context.

The Position Paper therefore establishes a principle central to Avallain’s approach: AI should assist professional users rather than replace professional judgement. Across our products, Avallain Author, Avallain Magnet and TeacherMatic, AI-powered functionality supports clearly defined educational tasks rather than autonomously determining grades, progression, certification or other learner outcomes. Professional review, adaptation and validation remain essential.

This matters because responsible AI cannot be achieved through technical safeguards alone. Nor can principles around ethics and human oversight be meaningful unless they are reflected in how systems are actually designed and operated.

Responsible AI therefore requires both a clear ethical and governance foundation and practical safeguards that translate those principles into the technology itself.

From Avallain Lab: Two Connected Publications

The relationship between the documents is deliberate and explicit, and both initiatives are part of the work of Avallain Lab.

Founded in 2023, Avallain Lab is Avallain’s academic and pedagogical resource hub. It supports product development while contributing to the broader e-learning ecosystem through ethical, research-informed approaches to educational technology.

The Lab is led by Professor John Traxler, UNESCO Chair and Commonwealth of Learning Chair, Academic Director of Avallain Lab and Carles Vidal, MSc in Digital Education, Business Director of Avallain Lab. An advisory panel of leading experts further strengthens its work, including Professor Rose Luckin, Professor of Learner Centred Design at UCL, with extensive expertise in AI in education.

Positioning these publications within Avallain Lab reflects the purpose behind both: to connect academic and pedagogical expertise, responsible technology design and practical guidance for publishers, institutions, educators and the wider educational community.

‘Responsible AI in education requires more than adopting powerful technology. It requires us to connect pedagogical purpose, human oversight and robust safeguards from the outset. Through Avallain Lab, these publications make Avallain’s commitment to the educational community more transparent: one sets out the principles that guide our use of AI, while the other shows how those principles are translated into concrete protections for data and users. Together, they provide a practical foundation for continuing to develop AI that supports educators and institutions without compromising human agency, safety or trust.’

Carles Vidal, MSc in Digital Education, Business Director of Avallain Lab

The Avallain Position Paper on AI Ethics and Security in Education provides the broader governance perspective. Written for educators, institutions, publishers and decision-makers, it sets out Avallain’s approach using European principles for trustworthy, human-centred AI as a baseline, including the EU AI Act, the EU Ethics Guidelines for Trustworthy AI and GDPR.

It addresses where Avallain’s responsibility lies across the AI value chain, how to consider risk, why human oversight is essential in educational contexts and how organisations should approach transparency, bias, data protection, learner safety and residual risk.

The Avallain AI Trust Model, in turn, focuses specifically on Avallain’s personal data protection strategy within generative AI workflows. It describes a four-layer safeguard framework that covers the journey of data through Avallain’s AI-enabled applications and features, from input to output.

At the User Application layer, secure interfaces, active prompt hygiene and AI literacy support safer interaction from the point where data is entered. The Sovereignty Layer introduces controls including regional data governance, personally identifiable information masking, identity decoupling, audit trails and data minimisation. Secure Transport protects data through encryption and regional endpoint controls, while the AI Inference Engine layer is designed around stateless processing, contractual zero-retention and restrictions that prevent user data from being used to train external AI models.

The accompanying Trust Model infographic reinforces this layered approach, showing privacy and security not as a single control applied at the end of a process, but as protections distributed across the full generative AI workflow.

The connection between the publications also works in both directions. The Trust Model explicitly places its technical safeguards within Avallain’s broader commitment to responsible AI and directs readers to the Position Paper for the wider ethical and governance context. The Position Paper, in turn, points to the Trust Model for the technical measures underpinning Avallain’s privacy and data protection commitments.

One establishes why and under what principles AI should be used. The other explains how key privacy and security principles are translated into technical safeguards.

What This Means for Avallain Intelligence

These publications also further reflect Avallain Intelligence, our framework for the responsible integration of AI in education, with ethics and safety at its core and a human-centred approach throughout.

Avallain Intelligence starts from the principle that AI should expand what educators, publishers and institutions can achieve without transferring professional authority to automated systems.

But human-centred AI means more than keeping a person ‘in the loop’.

Users need to know when AI is being used and understand that generated outputs are probabilistic, not authoritative. Institutions need the ability to decide where and how AI functionality is enabled. Educators need sufficient AI literacy to recognise limitations and review generated material appropriately. Learners need protection from inappropriate or unmediated uses of generative systems.

The data underpinning those interactions must also be handled accordingly.

This is where the AI Trust Model adds another practical dimension to Avallain Intelligence. Human-centred AI depends on protecting the people using it, and that includes protecting their privacy, their data and the content entrusted to educational systems.

Measures such as data minimisation, regional data sovereignty, PII masking, encryption, stateless processing and restrictions on external AI training translate those principles into infrastructure and operational controls.

Together, the publications show responsible AI as a combination of pedagogical purpose, human oversight, ethical safeguards, transparent governance and secure technical architecture.

That combination is particularly important in education, where trust cannot be separated from responsibility towards learners, educators, publishers and institutions.

Shared Responsibility Across the Educational AI Ecosystem

Another important conclusion from the Position Paper is that responsibility for AI does not sit with a single organisation.

Avallain acts as an AI system provider. It designs, integrates and operates educational applications built on third-party AI models, without developing or training the underlying models itself. Avallain is responsible for how AI capabilities are embedded into its products, how they are presented to users and what safeguards apply at the application level.

Model providers, technology providers, clients, institutions and professional users consequently hold different responsibilities within the same value chain.

Clients and institutions determine whether and how to enable particular AI features within their own organisational and regulatory contexts.

Educators and other professional users remain responsible for reviewing, adapting and validating AI-generated outputs before using those outputs in teaching and learning.

This shared-responsibility model is significant for publishers and educational organisations evaluating AI technologies. Responsible adoption cannot be reduced to selecting a technology provider. It also requires appropriate governance, staff understanding, clearly defined use cases and professional oversight within the adopting organisation itself.

The role of an education technology provider is therefore not to remove that responsibility, but to create systems that enable responsible practice.

What Comes Next: Responsible AI as an Ongoing Process

Neither publication presents responsible AI as a challenge that can be addressed once and considered complete.

Generative AI technologies continue to evolve. Regulation develops alongside them. Educational practices change as institutions and professionals gain experience with new tools. Risks that appear manageable today may need to be reconsidered as capabilities and use cases change.

The Position Paper therefore commits to continuous monitoring and improvement, drawing on ongoing research, systematic testing, collaboration with Avallain product teams, external pilots with educators and schools, and partnerships with researchers and experts in digital education and AI. Ethical considerations, user feedback and evolving regulatory expectations are intended to feed back into product design, user guidance and recommendations for the wider educational community.

The work of Avallain Lab is an important part of that process, providing the academic and pedagogical perspective needed to examine new developments critically and connect technological innovation with research, educational practice and responsible design.

The AI Trust Model follows the same principle at the infrastructure level. Its safeguards are designed around current requirements for secure generative AI workflows while allowing data sovereignty arrangements to respond to different regional and organisational requirements.

The papers do not set out a fixed product roadmap. Instead, they establish the direction in which Avallain intends to continue working: evaluating risk, strengthening safeguards, learning from real educational use and adapting its technology, guidance and recommendations as AI itself evolves.

For our clients, this provides a more transparent basis for understanding the principles behind Avallain’s AI-enabled products and the controls supporting them.

For educators and institutions, it reinforces the importance of purposeful adoption, AI literacy and continued professional oversight.

For publishers, it provides a clearer framework for considering how AI can support content and product workflows while protecting data, intellectual property and editorial responsibility.

Finally, for the wider educational technology sector, it reflects a broader requirement: innovation and responsibility cannot be treated as separate stages of AI adoption. They have to develop together.

Read the Avallain Position Paper on AI Ethics and Security in Education

The Avallain Position Paper on AI Ethics and Security in Education provides the broader context for Avallain’s approach to trustworthy, human-centred AI.

It explores human oversight, professional responsibility, risk classification, transparency, ethics and bias, information security, privacy, learner protection and the limitations of generative AI, offering practical context for organisations considering how to introduce and govern AI in education.

Explore the Avallain AI Trust Model

The Avallain AI Trust Model examines how personal data is protected throughout generative AI workflows, from the user interface and regional governance layer through secure data transport and AI inference.

It provides a closer look at the technical and operational safeguards behind Avallain’s approach to data protection, including data sovereignty, PII masking, encryption, stateless processing, zero-retention and restrictions on the use of customer data for external AI model training.

About Avallain

For more than two decades, Avallain has enabled publishers, institutions and educators to create and deliver world-class digital education products and programmes. Our award-winning solutions include Avallain Author, an AI-powered authoring tool, Avallain Magnet, a peerless LMS with integrated AI, and TeacherMatic, a ready-to-use AI toolkit created for and refined by educators.

Our technology meets the highest standards with accessibility and human-centred design at its core. Through Avallain Intelligence, our framework for the responsible use of AI in education, we empower our clients to unlock AI’s full potential, applied ethically and safely. Avallain is ISO/IEC 27001:2022 and SOC 2 Type 2 certified and a participant in the United Nations Global Compact.

_ 

Contact:

Daniel Seuling

VP Client Relations & Marketing

dseuling@avallain.com

Is research simply the pursuit of knowledge, or is there more to it? How does funding shape its priorities? How might different cultural contexts shape our understanding of its purpose, and how is AI changing the way we approach it? What does all this mean for the edtech industry? In this article, Prof. John Traxler explores the purpose of research, questioning how knowledge is created, whose interests it serves and what it means for the future of education and educational technology.

What’s the Purpose of Research?

Author: Prof John Traxler, UNESCO Chair, Commonwealth of Learning Chair and Academic Director of the Avallain Lab

St. Gallen, September 25, 2026 – This blog echoes our earlier one titled ‘What’s the Purpose of Education?’, adopting similar perspectives and raising similar questions about the role and impact of AI. Education and research might both be understood as the pursuit of knowledge, though in both cases this is not quite the case, even if there were any clear consensus about what actually constituted ‘knowledge’.

For many people, research is often assumed to mean the gathering of information, facts and data. It does, however, also embrace something more theoretical, abstract or conceptual, something else that might constitute knowledge; not merely observing and recording the falling of apples from trees but reflecting on what these observations might mean, developing an understanding of gravity and using that understanding to predict the behaviour of other material bodies.

Defining Research

Several ideas spring from this more abstract or conceptual view of research. One is that the conclusions are somehow durable and general; it is not just about apples, and it is not just today. Another is that research implies re-search, searching again, for findings that will challenge or refine or refute earlier conclusions, as illustrated by Eddington’s observations of the 1919 solar eclipse, which provided evidence supporting Einstein’s refinement of Newtonian gravity. In that sense, the knowledge generated by research is only ever provisional; good until there’s better, durable but not permanent.

Research in this conceptual or abstract sense, however, embraces research on research itself, not just a search for better data and better theories but also a search for a better understanding of theories themselves, of concepts, of abstractions, of knowledge, questioning the questions. 

This is sometimes portrayed as a greater criticality, questioning the meanings, language and methods of research, as well as the wider context in which it takes place. It also means considering researchers themselves; their motives, their options, their beliefs, their personalities, their backgrounds and their communities. We might ask, for example, why Newton had so much free time for speculation while so many of his contemporaries had so little, labouring perhaps as gardeners. Had their positions been reversed, what might their contributions to knowledge have been? Perhaps advances in fructivorous arboriculture rather than celestial mechanics.

The Modern Funding Landscape

Some of the more critical and contemporary questions about research continue to concern this wider context: who is funding it, and why are they funding it? Who benefits from it? Who loses from it? What is making it happen, or not happen? These questions bring us to the present day, prompting us to consider the arms trade, big pharma, market research and AI techbros; each funds vast amounts of research and raises questions about its purpose and motives.

Alongside this corporate research, there is also public funding for research, mostly in our universities. In our previous blog, we explored the ideal of education as a means of developing well-rounded individuals, drawing on the concepts of Bildung and the ‘renaissance man’. A similar ideal might once have been associated with university research: well-spoken, well-educated middle-class white men pursuing what we now call ‘blue skies’ or ‘curiosity-driven’ research. But if there was ever such a golden age, those days are over.

Public funding of research in many countries is now about national competitiveness, accounting for public money, bibliometric data such as citations, ‘taking ideas to market ’, and producing measurable ‘impact’. It is also about competition between individual researchers and their universities, encapsulated, for example, in the UK’s Research Excellence Framework. 

These days, the purpose of research is whatever the funder says it is. As has been frequently remarked, we live in an era of ‘policy-based evidence formulation’, not ‘evidence-based policy formulation’, in which ‘curiosity-driven’ or ‘blue-sky’ research has no purpose.

One consequence of this research environment is that no one wants to back losers; funders do not want to be responsible for wasting public money, so riskier research questions struggle to get answered, and so do the uncomfortable ones. Whilst a complementary perspective holds that, in large commercially funded research projects, ‘these are doomed to succeed’; they can’t go wrong because of the prestige invested in them.

Utopian, Dystopian and Different Cultural Perspectives

These depictions lead to competing utopian and dystopian views of the purposes of research, comparable to those earlier ones of the purposes of education. The utopian may be the Olympian pursuit of knowledge for its own sake and the greater good of humanity, undertaken by some noble elite. The dystopian, by contrast, is more likely the rush to fund any question that leads to some market advantage. 

This, though, should encourage us to ask whether these perspectives are universal, or are they shaped by particular cultural and intellectual traditions? Equally, we should not assume that other traditions and cultures don’t share these ideas. Do these purposes look different elsewhere, away from globalised universities, organisations and institutions? Does the question of purpose take on a different meaning or is it even relevant in every context? Are ideas such as pursuing, questioning and extending knowledge shared?  

Communities may understand, create, develop and share knowledge in different ways, encompassing the physical and social worlds as well as, in some traditions, the natural and spiritual. Rather than attempting to characterise these knowledge systems collectively, we might ask how different traditions broaden or challenge our understanding of what research is and what purposes it serves. 

We might also consider what happens when knowledge traditions with different levels of institutional, economic or political power encounter one another. This raises questions about who benefits from knowledge and research, who controls it and how knowledge originating within particular communities is recognised and used.

The Role and Impact of AI in Research

So what is AI’s impact on research? Decades ago, I was told that exploiting a technical innovation happens in three phases: firstly, it is used to solve existing previously difficult problems; then to solve previously impossible ones; and finally to solve previously inconceivable ones. Recent workshops on the impact of AI on research suggest that we are mostly at the first stage, merely making existing research procedures and practices happen quicker, bigger, sooner and cheaper. Researchers are mostly just reacting to stay in the game and exploiting these new tools, though, of course, ‘A rising tide lifts all boats’, the boats being researchers’ competitors. 

With tools like Anthropic’s Interviewer, which can interview thousands of people in days more competently than many human social scientists can in months, we are clearly entering the second phase. So how should researchers be addressing the third phase? When this happens, will it be changing the nature of knowledge, not merely rendering some of it obsolete but changing something fundamental? At this stage, it is, of course, impossible to conceive of the inconceivable, bringing to mind the Rumsfeld Matrix with its ‘unknown unknowns’. 

Finally, what is the purpose of research for the edtech industry? It sits at the intersection of the purpose of education and the purpose of research and is a fluid and unstable compound of all the features of both, utopian and dystopian. 

About Avallain

For more than two decades, Avallain has enabled publishers, institutions and educators to create and deliver world-class digital education products and programmes. Our award-winning solutions include Avallain Author, an AI-powered authoring tool, Avallain Magnet, a peerless LMS with integrated AI, and TeacherMatic, a ready-to-use AI toolkit created for and refined by educators.

Our technology meets the highest standards with accessibility and human-centred design at its core. Through Avallain Intelligence, our framework for the responsible use of AI in education, we empower our clients to unlock AI’s full potential, applied ethically and safely. Avallain is ISO/IEC 27001:2022 and SOC 2 Type 2 certified and a participant in the United Nations Global Compact.

Find out more at avallain.com

_

Contact:

Daniel Seuling

VP Client Relations & Marketing

dseuling@avallain.com