Generative artificial intelligence has swept into classrooms around the world faster than most education systems can write rules for it, and nowhere is the gap more striking than in the years before school begins. A new qualitative study published in the journal AI, Brain and Child examines how Australian educational experts think generative AI should be handled in early childhood education and care policy, and its findings paint a picture of broad support for policy action wrapped in unusually heavy caution.
The research, conducted by Luyao Liang and Rongle Tan and published on 10 August 2026, drew on in-depth interviews with seven Australian academics who had expertise in early childhood education and care, educational technology, or generative AI in education. Using purposive and snowball sampling from an initial pool of thirteen candidates, the team recruited participants ranging from junior to senior academic positions and interviewed them between October and November 2024 in sessions lasting 34 to 57 minutes, averaging around 43 minutes. The transcripts were then analysed through a six-step thematic analysis following established qualitative guidelines, with peer debriefing, an audit trail, and thick description used to strengthen trustworthiness.
The study’s starting point is a documented policy imbalance. Australia has an explicit national instrument, the Australian Framework for Generative Artificial Intelligence in Schools, that guides ethical and responsible use of the technology in school settings. Early childhood education and care, by contrast, is covered only by broader digital frameworks: the Early Years Learning Framework and the National Quality Framework acknowledge that digital technologies are ubiquitous in children’s lives and stress intentional, educator-mediated use, but neither explicitly addresses generative AI. The most sector-specific guidance currently available is Early Childhood Australia’s revised 2025 Statement on young children and digital technologies, which names AI applications and addresses issues such as conversational agents, privacy and consent risks, and data harvesting by AI-enabled toys. Yet the authors note that this Statement remains an advisory professional document rather than a formal policy instrument, and its scope is broader than generative AI alone.
Against that backdrop, the interviews produced a clear headline result: most of the experts, five out of seven, advocated strongly for integrating generative AI into early childhood policy while attaching important conditions and uncertainties. One participant offered conditional support contingent on sector-based empirical research and stakeholder engagement, and one expressed uncertainty about whether such policy should be developed at all given the pace of technological change. Notably, no expert categorically opposed policy development, which the authors read as a broad consensus that policy engagement with generative AI in early childhood settings is necessary. Crucially, however, the support was cautious, protective, and conditional rather than enthusiastic.
The rationales behind that support clustered into three themes. Child welfare dominated. Experts called for national and institutional regulation to safeguard children’s safety, security, and privacy, with specific attention to how sensitive data is collected, stored, and used; one expert warned of the possibility of a systemic data breach harming children and families, and another noted that many childcare centres operate insecure websites. Developmental appropriateness formed a second strand: some experts cited young children’s developmental vulnerability, including concerns about screen addiction and associated language delays, as grounds for heightened caution, and one questioned the necessity of generative AI policy at this stage at all, arguing that children are naturally curious without technological intervention and that there is plenty of time to engage with the digital world once they reach primary school. Ethics formed the third strand, with experts describing a regulatory and ethical void in which there are no guardrails or guidelines for how such tools should and should not be used in teaching and learning.
Professionalism in early childhood education and care emerged as the second major theme. Experts worried that relying on generative AI to automate documentation, the observation, documentation, and reflection cycle at the heart of early childhood pedagogy, would bypass the cognitive effort that makes that work meaningful, and they flagged the risk of educators relying on fabricated or low-quality AI-generated records for core professional tasks. Concerns were also raised that over-reliance on the technology could erode teachers’ capacity to critically evaluate its accuracy, and several experts questioned whether generative AI is compatible with the play-based learning that defines the sector. One participant highlighted the absence of a clear theoretical foundation, noting that early childhood education is theorised around children’s learning and play while previous research has tended to treat technology merely as a tool.
The third theme concerned structural pitfalls in the policymaking process itself. Experts described delayed policymaking, citing a historical example of sector technology standards that arrived years too late, and warned that a similar lag would widen the disconnect between early childhood settings and schools. They criticised generic policy that treats digital technologies as a homogeneous category without distinguishing the distinct capabilities and risks of generative AI from earlier tools such as tablets or cameras, leaving educators unsure where to start. They also criticised top-down policymaking perceived as disconnected from the realities of practice, with one expert describing current efforts as superficial exercises that prioritise symbolic acknowledgment over practical guidance.
On what policy should contain and how it should be built, the experts converged on a blended top-down and bottom-up model. They envisioned an overarching governmental framework that establishes foundational requirements while allowing adaptation at the individual service level, informed by grassroots practitioner knowledge and translated into practical guidance by professional bodies such as Early Childhood Australia. They identified a wide range of stakeholders who should be involved: academic experts across areas including online safety, Indigenous education, inclusive education, and culturally diverse perspectives; practitioners across the full diversity of roles and qualifications in the workforce; government bodies and institutions; and, more contentiously, children and the technology industry. Views on children’s direct participation diverged, with some experts valuing children’s voices in principle while acknowledging the practical difficulty of eliciting them. Views on industry involvement ranged from outright exclusion, on the grounds that companies selling products face a conflict of interest with children’s welfare, to cautious inclusion under explicit sector-set guidance.
The experts were also candid about the obstacles ahead. They identified insufficient understanding of generative AI among potential policymakers, a pace of technological change that outstrips current knowledge, and a scarcity of research on what effective use of AI in early education actually looks like. Inadequate resource allocation posed a further barrier: expert consultation requires budgets that under-resourced early childhood services and agencies may lack, in a sector that receives fewer government resources than schools. Even if policy is written, experts worried about uptake, warning that a framework could become another document on the shelf, symbolic compliance rather than substantive practice, and stressed the need for consistency across the National Quality Framework, the Early Years Learning Framework, and professional statements.
The authors argue that generative AI policy for early childhood should not simply extend school-based frameworks downward but should be sector-sensitive, participatory, enactable, and aligned with Australia’s existing early childhood frameworks. They suggest one possible mechanism: requiring documented, framework-aligned generative AI use policies at the service level, assessed through existing quality-rating processes, with the Early Years Learning Framework supplying overarching principles and professional bodies translating them into practice, though they stress this is a starting point requiring substantial consultation and piloting. The study’s limitations are explicit: the participants were academics rather than frontline educators, service directors, families, or children, and no policymakers responsible for implementation were included, which the authors acknowledge likely biased findings toward protective, academically reasoned positions and away from the practical trade-offs practitioners navigate daily. As young children increasingly encounter conversational agents and intelligent devices at home and in educational settings, the study concludes that the early childhood sector can no longer remain peripheral to debates about governing generative AI in education.
Subject of Research: Expert perspectives on integrating generative AI into Australian early childhood education and care policy
Article Title: The integration of generative artificial intelligence into early childhood education and care policies in Australia: an expert interview
Article References: Liang, L., & Tan, R. (2026). The integration of generative artificial intelligence into early childhood education and care policies in Australia: an expert interview. AI, Brain and Child, 2(1), Article 16. https://doi.org/10.1007/s44436-026-00041-7
Image Credits: AI Generated
DOI: 10.1007/s44436-026-00041-7
Keywords: generative AI, early childhood education and care, education policy, Australia, expert interviews, thematic analysis, child welfare, developmental appropriateness, educator professionalism, policymaking, stakeholder engagement, data privacy
News Source: Courtney Benton. (October 28, 2026). Australian experts urge cautious, sector-specific AI policy for early childhood. Scienmag.
