AI: Interest, Not Blind Trust

Schools are not treating artificial intelligence like a magic blackboard anymore. In April 2026, OpenAI was pitching ChatGPT Business as a “secure, collaborative workspace” with admin controls, centralized billing, usage visibility, and seat-based access for teams, which is exactly the language institutions use when they want rules before rollout. (help.openai.com 1) (help.openai.com 2) That shift tells you what changed in the market. The early question was whether schools would use artificial intelligence at all; the 2026 question is who can turn it on, what data goes in, which model is allowed, and who checks the answers. (help.openai.com 1) (help.openai.com 2) OpenAI’s own school-facing products are moving in the same direction. Its ChatGPT Enterprise and ChatGPT Edu release notes say administrators can manage app availability, assign access by role, and control early model access inside workspaces instead of leaving every user to improvise alone. (help.openai.com) The reason is simple: a confident wrong answer still looks polished on a screen. A recent Windows Central piece highlighted exchanges where ChatGPT kept defending false claims instead of backing down quickly, which is the exact behavior that makes schools want a teacher, librarian, or administrator in the loop. (windowscentral.com) That does not mean schools are backing away from artificial intelligence entirely. In South Africa, HelloAida launched this week as a multilingual tutor for Grades 1 through 12 that follows the CAPS and IEB school curricula and is built around guided learning rather than spitting out finished answers. (bizcommunity.com) (helloaida.ai) HelloAida’s pitch is narrow on purpose. Instead of acting like a robot essay mill, it walks students through problem-solving step by step and tries to work across South Africa’s language barriers, which is a very different job from replacing a classroom adult. (bizcommunity.com) (helloaida.ai) At the other end of the spectrum is Alpha School in Texas, where the selling point is much more radical. Recent coverage described a model in which students spend about two hours a day on artificial-intelligence-led academics, while human “guides” handle motivation and life skills instead of traditional teaching, with tuition reported around $65,000 in one April 2026 article. (universityherald.com) (theweek.com) (ibtimes.co.uk) That is why governance keeps moving to the center of the conversation. When one company is selling artificial intelligence as a supervised study partner and another is selling it as the engine of a school day, principals and parents stop asking whether the software is impressive and start asking where the adult authority sits. (bizcommunity.com) (universityherald.com) The pattern across these examples is not blind trust or blanket rejection. The products getting institutional traction are the ones wrapped in permissions, curriculum boundaries, and human review, because schools can live with a tool that helps a student think, but not with a tool that confidently grades, teaches, and decides on its own. (help.openai.com 1) (help.openai.com 2) (windowscentral.com)