The Hidden Reality of AI in Schools: Risks and Opportunities

01What the Data Reveals About Student AI Usage

New research paints a nuanced picture of artificial intelligence in schools: students are using these tools far more than their parents realize, and a significant share of those interactions raise serious concerns. For school leaders, the challenge has shifted from whether to allow AI to how to govern it responsibly. An analysis of 1.2 million student conversations with AI, conducted by school safety platform Securly across 1,312 districts in 39 U.S. states, found that 20% of interactions triggered content flags. Of those flagged conversations, 94.6% involved students attempting to have AI complete their assignments outright. More alarming still, approximately 2% of all prompts showed signs of self-harm, bullying, or violence — representing more than 24,000 potential crisis moments buried inside chatbot conversations that most districts had no visibility into. These findings align with a February 2026 Pew Research Center survey of 1,458 U.S. teens ages 13 to 17, which found that 54% have used chatbots to help with schoolwork, and one in ten report doing all or most of their schoolwork with AI assistance. Some 59% of teens believe that using AI to cheat happens at least somewhat often at their school. In a separate survey, Morgan Polikoff of USC Rossier School of Education found that only 7% of parents believed their teens used AI for schoolwork multiple times a week, while 27% of teens reported doing so — and that figure, Polikoff notes in an EdSource commentary, is likely an undercount. The concern extends well beyond academic integrity. A Brookings Institution report warns that the technology market is outpacing scientific research, putting child development, learner autonomy, and data privacy at risk. Researchers note that young children have no meaningful ability to understand or consent to the data practices of the AI products they interact with — a gap that existing regulation has not yet closed.

02The Brazilian Response: MEC's New Guidelines

Brazil's federal government has moved to address these challenges directly. On April 8, 2026, the Ministry of Education (MEC) launched the guiding document "Artificial Intelligence in Basic Education", developed in partnership with UNESCO. The document establishes ethical, critical, and safe integration of AI in schools, aligned with the National Common Curricular Base (BNCC) and Brazil's General Data Protection Law (LGPD). The document is organized around two complementary goals: teaching *about* AI and teaching *with* AI. It also introduces a free five-module training course for high school teachers — "AI in Teaching Practice: Ethical, Creative, and Pedagogical Use" — designed to help educators critically analyze and responsibly integrate these tools into their classrooms. Secretary of Basic Education Kátia Schweickardt framed the initiative plainly: "Artificial intelligence is already a reality, it is no longer a choice. We must not be afraid, we must learn to use it." The Brazilian approach reflects a broader consensus emerging from the data: outright bans are ineffective. As the Securly analysis demonstrates, blocking AI access on school networks simply pushes student activity off-network and out of institutional sight. A Texas district that instead adopted a monitoring-and-redirection model saw weekly deflections to unapproved AI tools drop by 90% in the first week, while maintaining nearly 25,000 educational AI conversations per week within a supervised framework.

03From Research to Practice

The evidence available in April 2026 points toward a clear, if demanding, path forward. AI in education is a powerful tool that requires pedagogical intentionality — not prohibition, and not uncritical adoption. The data on cheating and risky behavior is not a reason for alarm, but a mandate for action. Teacher preparation is the most consequential lever available to school systems. Educators who understand how AI works — and where it fails — are better positioned to design assignments that resist superficial AI completion, to recognize when a student's AI interaction signals distress, and to model the kind of critical engagement with technology that students will need throughout their lives. When AI handles repetitive administrative tasks, it frees teachers for the relational work that no algorithm can replicate. Digital and media literacy must also become a core curricular priority. Students who learn to question AI-generated answers, identify algorithmic bias, and understand the limits of these systems are not just safer users of the technology — they are better thinkers. Learning is a process, not merely a product, and any educational use of AI that shortcircuits that process undermines the very goal it claims to serve.

04What to Watch Next

The question facing school systems worldwide is no longer whether students will use AI, but whether institutions will have the visibility and the frameworks to guide that use productively. Brazil's MEC guidelines offer a meaningful starting point, but their impact will depend on how school networks implement them and how consistently teachers receive the support they need. In the months ahead, the most important indicator to watch is not adoption rates, but learning outcomes: are students developing deeper understanding, or simply producing more polished outputs?