Google hands schools the AI switch as learning concerns grow

Google said on Tuesday it would turn off access to Gemini, Gemini Notebook, Gemini in Classroom and Google Search by default for school-managed K-12 student accounts, requiring administrators to explicitly enable the services.

The company said the change followed feedback from schools, teachers, researchers and policymakers seeking greater control over classroom AI use. It did not announce a rollout date, saying it would share details with schools in the coming months.

Google is also expanding Guided Learning, which helps students work through problems step by step. It describes the approach as “productive struggle,” where pupils are challenged enough to build problem-solving skills without becoming overwhelmed.


“We’re building Guided Learning capabilities into Search for student accounts to help them work through problems without skipping straight to answers. When our systems understand a student is looking for certain types of homework help, Search will walk them through the topic or problem step by step,” Google wrote.

Guided Learning became available globally in Google Classroom on Tuesday. The Search version will begin with a pilot in select US schools later this month before expanding to other school districts.

The distinction between getting help with an answer and learning how to reach it is central to an emerging body of research.

In a study of nearly 1,000 high school mathematics students in Turkey, researchers found that students' grades improved 48% when they could use ChatGPT to solve problems. However, when access to the chatbot was removed, these students performed 17% worse than those who had never used the tool.

A study published by the American Psychological Association (APA) discussed cognitive offloading, where people use external tools to reduce the mental effort needed to complete a task.

“If your goal is learning, then efficiency and ease should take a back seat,” Brooke Macnamara, an associate professor of psychological sciences at Purdue University, told the APA.

In India, a survey of low-income households points to frequent use of generative AI alongside gaps in understanding the technology.

Central Square Foundation’s Bharat Survey for EdTech 2025 found that 35% of children who already used educational technology were using generative AI for learning. Conducted between July 2025 and January 2026, the survey covered 12,500 households and 2,500 teachers across 10 states. Among children using generative AI, 96% used it multiple times a week and 69% used it daily.

Among generative AI users, 73% sought help with doubts and practice, 48% with new skills, language learning or translation and 32% with test preparation. Easy-to-understand explanations, quick responses and interactive formats were among the main attractions.

UNICEF said in June that an analysis of data from 10 countries estimated at least 20 million children in those countries had used AI, including 13 million for learning and homework and more than two million for advice about things that worried them.

Evidence about AI’s role in cognitive development and emotional dependency was still emerging, UNICEF said, while highlighting risks involving misinformation, privacy and manipulated images.