Google recently announced that its AI tools, including Gemini and Gemini Notebook, would be turned off by default for kindergarten to Grade 12 student accounts starting in early 2027. Google Search, its popular search engine, would also be turned off by default.The New York Times reported that Chris Phillips, vice president for education at Google, said: “Search and A.I. are deeply integrated,” and that if a school has banned A.I., “they will not have access to search.”This means schools that want students to simply search on Google will have to accept Google’s AI whether they want it or not. As researchers who study education policy and technology, we see Google’s move as a wake-up call. It exposes how much power one company holds over schools, and why Canadian governments working with school boards, teacher unions and parents need to set the terms for how AI enters classrooms. Setting common regulatory standards at the federal and provincial levels could give schools the collective leverage they currently lack when dealing with trillion-dollar companies like Google. Read more: The release of AI chatbots has harmed teens’ learning and Canada needs to act How did we get here?Google presents the change as giving schools more control amid pushback from child-safety advocates, parents and educators after it switched on Gemini by default in Google Classroom for many K-12 students in August. Google Classroom is an online platform where teachers post assignments and students submit their work.Google has also said AI in its search function will guide students through problems step by step rather than give them quick answers. But guided or not, AI remains built in.Over the past 15 years, Google has become one of the largest providers of digital infrastructure to schools around the world largely by offering its tools for free or at low cost — although researchers highlight hidden costs in the form of data trails left exposed by privacy loopholes.Google’s ‘ecosystem’Its influence comes not simply from selling individual products to schools, but from embedding schools within an interconnected ecosystem of software applications such as Google Classroom, hardware like Chromebooks and cloud-based services like Google Drive — all increasingly relied upon by schools. About 150 million students and teachers worldwide now use Google Classroom while 60 per cent of all Chromebook purchases are for use in education.This ecosystem approach gives Google immense leverage over schools. Even when offered a “choice,” schools cannot simply switch away from Google overnight. Switching to alternatives involves not only purchasing new software and hardware but also migrating data, retraining teachers and students and disrupting established routines and practices. Google’s decision to tie Search to AI shows how this dependence can become a form of coercion.‘Pipeline of future users’Although Google presents its investments in education as altruistic and benign, its commercial interests have always been central to its strategy. Internal company documents, filed in a U.S. lawsuit against Google and other social media companies, describe Google’s work in schools as building a “pipeline of future users.” A Google spokesperson said the documents “mischaracterize our work.” Read more: We street-proof our kids. Why aren’t we data-proofing them? Google Search, the company’s biggest revenue-generating business, lies at the core of that pipeline. As AI chatbots like OpenAI’s ChatGPT and Anthropic’s Claude gain popularity as alternatives to traditional internet search, they threaten not only Google’s current user base but also its future one. A generation that learns to access information online without “Googling” as its primary gateway is what Google wants to prevent. Seen from this lens, it’s no surprise Google isn’t letting schools search on Google without AI features embedded in one form or another. Who is setting the rules?This conundrum facing schools raises a central question: who sets the direction and conditions for AI in K-12 education?The Council of Ontario Directors of Education’s 2026 “AI in Education” resource cautions that students and staff are already using AI “in the absence of guidance.” Read more: Should AI chatbots simulate care for students? Alberta teachers say no For the 2026-27 school year, the Ontario education ministry required “Cyber Safety and Artificial Intelligence” to be one of four mandatory topics on teachers’ professional activity days, which includes getting familiar with AI tools. British Columbia has published a set of considerations to help school districts select AI tools. But it sets no rules of its own, leaving individual districts to make their own decisions. This approach assumes districts have meaningful choices. But Google sets the same terms for all schools, and changes them when it chooses.At the federal level, the emphasis has been more on AI adoption. In September, the federal government launched a $13-million National AI Literacy Initiative, in partnership with the Alberta Machine Intelligence Institute, that includes free, self-paced modules that show K-12 educators how to integrate AI into curriculum, assessment and teaching. The Canadian Teachers’ Federation has warned that the initiative puts “adoption before safety,” encouraging teachers to bring AI into classrooms before its safety and educational benefits can be established.Regulatory standards neededGoogle’s ultimatum shows why a human-centred approach to AI in education must begin with a fundamental question: Who gets to decide whether and how AI enters classrooms? Such an approach prioritizes students’ needs, rights and learning and teachers’ professional judgment over technology companies’ commercial interests.UNESCO’s guidance on generative AI in education points in this direction, setting out distinct responsibilities for governments, AI providers and users. Researchers have argued AI should be regulated through a developmental lens before it enters K-12 classrooms, and there is already evidence of why such safeguards are warranted. Read more: School board social media lawsuits: For too long we’ve sought individual solutions to a collective problem In July 2026, Common Sense Media gave the AI features in Google Search its lowest possible rating, “unacceptable risk,” after finding that they completed students’ homework assignments and failed to respond appropriately to children in crisis. Google disputed the findings, saying the testing relied on “a narrow set of ambiguous and contrived queries.”Questions about whether these tools are safe for children shouldn’t be settled by school boards alone.Setting joint standards allows school boards to weigh the educational value of AI against other pressing priorities, rather than treating its adoption as inevitable in already underfunded education systems.Rajender Singh is a Senior Policy Analyst at The Dais, a university based think tank at Toronto Metropolitan University, whose work includes technology and K-12 education.Melody Viczko and Yadira Lizama Mué do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.