Google plans default-off Search and AI for school accounts

Google plans to switch off Google Search and several artificial intelligence services by default for K–12 student accounts, requiring school administrators to enable access. The changes are expected in early 2027, according to New York Times reporting republished by The Philadelphia Inquirer.

Schools will be able to enable individual products rather than accept the entire collection. But that distinction does not give districts an AI-free version of Google Search: the search service itself includes AI features, leaving schools that prohibit those features with a separate decision about whether to allow Search.

What exactly is Google changing?

In its official announcement, education chief Chris Phillips said Google was responding to feedback from educators, researchers and policymakers seeking greater school control. Administrators must explicitly enable the listed services.

The announcement concerns school-managed K–12 student accounts. It does not announce an internet-wide shutdown of Search or a ban on other search engines.

Can schools keep Search and turn off Gemini?

Yes, according to Google’s clarification, reproduced by Search Engine Roundtable. Google said administrators can enable one, two or all of the named services individually. The clarification also said AI Mode and AI Overviews in Search would offer step-by-step learning support.

The distinction matters because disabling a standalone chatbot and banning all generative AI are different policies. A school could keep Gemini unavailable while permitting Google Search, including the AI functions within Search. A policy against every AI-generated search answer would pose a different problem.

Why is the Search decision drawing criticism?

The Associated Press reported objections from educators and policymakers, including former Federal Trade Commission chair Lina Khan, who accused Google of using its search-market position to pressure schools toward AI adoption. Children’s advocacy group Fairplay also criticized the approach and urged schools to consider other research tools.

Those are critics’ assessments of the company’s conduct, rather than findings established by the announcement. Google presents the change as an expansion of school choice and control.

The dispute has an unusually large audience. AP reported Google’s figure of 150 million students and teachers using Classroom worldwide. That number describes Classroom’s broader user base; it is not a count of K–12 students whose Search settings will change.

The underlying disagreement is about what constitutes a meaningful choice. A separate switch for an optional chatbot is straightforward. A decision about a widely used search service becomes more complicated when the service includes features that a district’s instructional policy excludes.

How did schools reach this point?

The New York Times reported that Google switched on AI features within Classroom in August for K–12 schools that had already allowed separate student access to Gemini. Some school leaders objected to the way that change reached classrooms. The planned default-off approach reverses that starting assumption.

The same report described two prospective Search experiences for school accounts: one with AI summaries and chat features, and another using Guided Learning. Neither was described as a return to Search without generative AI.

Education Week reported that settings could differ by grade. Google Classroom product lead Brittany Siegal Mennuti described different levels of access, including restricting one group, limiting another to Guided Learning and permitting older students a fuller search experience.

That creates practical work for administrators during the school year. They must connect technical settings with instructional rules: what younger pupils may use, what older students may use and what teachers can assign. An account setting alone does not explain an assignment’s rules to a student or parent.

Education Week also reported that some filtering approaches can hide AI content, but described concerns that smaller districts may lack staff for technical workarounds. Such approaches are distinct from an AI-free Search option supplied directly by Google.

What is Guided Learning supposed to do?

Google says its approach is to coach students through problems rather than let them skip directly to answers. Its Nov. 3 announcement says a Search pilot will begin at selected U.S. schools later in November, while Guided Learning in Classroom became available globally that day.

In a separate product and research announcement, Google described a four-week Classroom pilot with the Ark school network at three British secondary schools. Teachers sought better visibility into student interactions and ways to keep conversations focused on assigned materials.

Google says the resulting Classroom features allow teachers to assign learning experiences tied to course materials and receive information about students’ progress. The company reported that teachers in the pilot found the system could handle initial questions while they provided individual help elsewhere.

Those observations describe a particular pilot and feedback from its participants. They do not establish how every classroom will perform, whether students will consistently follow the intended workflow or how much additional oversight teachers will need.

Chalkbeat reported that teachers will be able to review student chats and that students will not be able to switch from the Classroom learning experience back into ordinary Gemini within Classroom. Google officials told the outlet that this version had stronger restrictions than the consumer tool.

The classroom version therefore should not be treated as identical to opening a general-purpose chatbot. The relevant tests include whether it stays with the lesson, whether teachers can understand what students did and whether students can demonstrate understanding afterward.

What does the evidence show about learning and safety?

Google’s Oct. 29 account of a British mathematics study offers evidence with important limits. The 10-week randomized trial involved more than 1,100 students ages 12 to 15, using the Eedi platform. Human experts checked every AI response before students received it.

Google reported that AI-supported students performed similarly to students receiving human tutoring on in-platform work and intermediate quizzes. However, the reported gains did not appear in independent STAR standardized-test scores during the trial.

That distinction matters when assessing claims of educational benefit. Progress on an activity within a tutoring platform is not the same measurement as broader achievement on a separate test. The study’s real-time human review also differs from unrestricted chatbot access. The results support examining specific classroom designs, rather than assuming all uses of AI produce the same outcomes.

Safety concerns predate this week’s announcement. In a July assessment, Common Sense Media’s Youth AI Safety Institute reported testing more than 2,600 searches with accounts configured for young users. It criticized inaccurate answers, inappropriate responses and the completion of schoolwork rather than assistance with learning.

The institute said AI Mode supplied complete responses to all 180 assignments it tested and recommended that Google make AI search features default-off for minors. Those findings concern the products and tests examined at that time. They are not a completed evaluation of the newly announced school Search pilot.

What are lawmakers asking Google?

The scrutiny also includes a bipartisan congressional inquiry. In an Oct. 21 letter announcement, Sens. Richard Blumenthal, D-Conn., and Josh Hawley, R-Mo., demanded answers from Alphabet and Google chief executive Sundar Pichai about AI’s introduction into classrooms.

They raised concerns about independent academic work, emotional development and the company’s knowledge of potential harms. Their inquiry puts questions about student safeguards alongside the narrower dispute over default settings. Allegations and concerns in a congressional letter are not, on their own, proof of causation or a legal finding against the company.

The wider debate over AI accountability also extends beyond Google, as NewsCubby’s coverage of state lawsuits involving OpenAI explains. Product changes are another part of that landscape, including recent changes to ChatGPT.

What do we know, and what remains unknown?

The announcement describes changes to access controls, while the classroom research examines particular uses of AI. Neither establishes that simply enabling a service will improve learning across an entire district.

What remains unresolved is how the final rollout will work for particular districts, how well the new learning experience will prevent shortcuts and whether its benefits will carry over to independent student performance. A product announcement cannot settle those questions. The pilot and later implementation will provide opportunities to test them.

What’s next?

Education Week’s reporting puts the next steps across two periods: a school pilot this month and broader changes early next year. That leaves administrators time to examine the available details before deciding how to implement their policies.

For school communities, the next consequential information will be the district’s own access decisions and explanations of which features students will encounter. Families will need that local information to know what the national announcement means for their children’s actual schoolwork.

*Want more? Explore today’s headlines on NewsCubby and join My Day free to make NewsCubby part of your day.*

Photo credits

Article, cover and Short: Google Headquarters Google Logo, photographed April 2, 2014, by Anthony Quintano. CC BY 2.0. Archive image of Google headquarters, not the 2026 announcement. Resized in this article; cropped, camera motion and text overlays added for the Short and cover.

Short: Samsung Chromebook, photographed July 6, 2011, by Closari. CC BY 2.0. Prior rotation and crop by Anna Frodesiak, February 4, 2012. Additional crop, camera motion and text overlays by NewsCubby. Archival hardware illustration, not a current product or a demonstration of the announced features.

Short: Google office logo in Toronto behind tree branches, photographed August 26, 2026, by Wiki.cullin. CC0 1.0 Universal Public Domain Dedication. Cropped, camera motion and text overlays added by NewsCubby. This is Google's Toronto office, not Mountain View headquarters or the announcement event.

The photographs retain their stated licenses. Company names and marks are used for editorial identification, without implying endorsement. These archive photographs do not depict students or a current classroom.

*NewsCubby articles are produced with AI assistance and reviewed by humans.*