With an increase in the use of Artificial Intelligence (AI) tools globally, including in sub-Saharan Africa, significant concerns have been raised regarding their impact on teaching and learning. This prevailing discourse often highlights the potential negative impact, such as the lack of trust in students' cognitive abilities and the authenticity of their work, and other ‘theoretical’ negative impacts. While these concerns remain plausible, they tend to overlook the benefits AI holds for creating educational opportunities, especially for children in vulnerable populations. Along this line, in this conceptual paper, we argue that AI tools, such as adaptive Learning Management Systems (LMS), can serve as an important educational lifeline for undocumented migrant children, particularly in contexts like South Africa where their access to basic education is largely restricted by systemic and social barriers.

South Africa's response to increased intra-regional migration has been highly securitised and nationalistic, fostering a climate of discrimination and xenophobia. Migrants, especially the undocumented, are often pejoratively labelled and seen as threats, creating a dilemma for providing basic services like education. Undocumented migrant children, many born in South Africa, are trapped in a legal paradox. While the Constitution and Schools Act guarantee a right to basic education for all, the Immigration Act effectively criminalizes the presence of an "illegal foreigner" in schools. The policy conflict, combined with administrative barriers like requiring birth certificates for admission, systematically excludes these children from their fundamental right to education and dignity.

This exclusion is compounded by several factors including lack of documentation is the primary barrier, xenophobic attacks affect children's educational capability, access to healthcare and other services is difficult, policy ambiguities persist, court rulings are not implemented, and Covid-19 further limited access. In this context of systemic failure, we argue AI-powered tools offer a significant alternative for equitable educational access. While the formal system “gatekeeps”, AI platforms can bypass these institutional barriers by providing context-based, personalised, remote learning, alternative educational pathways, and empowering students with agency.

To conclude, we posit that the perspective of AI and its powered tools’ role in education must be reframed to include its potential to serve and significance in practically serving marginalised children population in specific contexts. Therefore, we recommend a strategic focus towards the implementation of contextualised AI-driven educational solutions. While advocacy for systemic change remains important, we note that it cannot be the only strategy while the educational capabilities of children are actively being restrained. By providing such alternative educational pathways, there would be increased opportunities to move beyond physical institutional gatekeeping which perpetuates a cycle of exclusion. This is not a stop-gap measure but a workable intervention that allows for the upholding of the right to education, affirms the inherent dignity of every child regardless of their legal status, and leverages 21st-century technology to address a pressing human rights concern.

Keywords: Artificial Intelligence (AI), Education, Undocumented migrant children, South Africa.