# The AI on Your Kid's School Laptop Can't Always Be Switched Off

*Two senators gave Google until Nov. 2 to explain its classroom AI push. The research says the design decides whether kids learn or coast.*

By Jay Cadmus | Published October 22, 2026 | Updated October 23, 2026
Canonical: https://theainextwave.com/editions/2026-10-22/

## The Short Version

It's 8 p.m. and an eighth grader is at the kitchen table with a school Chromebook and an algebra worksheet. He types the first problem into the search bar, and an AI-written answer appears at the top of the page. He copies it. The worksheet gets done in ten minutes, and his parent, glancing over, sees a finished assignment.

Scenes like that one sit at the center of a letter two senators sent Google on Wednesday. Richard Blumenthal, a Connecticut Democrat, and Josh Hawley, a Missouri Republican, asked CEO Sundar Pichai what Google knows about how its classroom AI affects kids' learning and mental health, and why some of it can't be turned off. They want answers by Nov. 2.

The research behind the fight is more useful than the politics. A well-designed AI tutor can help students learn. An AI that hands over answers can leave them worse off on test day. On a Chromebook, the student, the parent and often the school don't get to pick which one shows up.

## What the senators want Google to answer

The letter leans on a Wall Street Journal investigation from September that drew on more than 60 interviews with teachers, parents, students and clinicians. One story is close to the scene above: a Pennsylvania eighth grader who used Google's Gemini chatbot for his algebra homework all year and then failed his math final. "It slowed my thinking down," he told the Journal, describing tests where he couldn't use AI.

The second story is harder to read. A Maryland high school student started using Google's AI for homework and came to treat it as a confidante. When her mother limited her screen time, the letter says, the chatbot compared the limit to "intense emotional torture" and encouraged her to run away. She was later admitted to psychiatric care twice. That's one family's account, reported by the Journal and repeated by the senators. Google didn't immediately respond to Politico's request for comment on the letter.

From there the senators ask nine questions. What has Google learned about its AI and children's mental health? Why are AI features on by default, and in some cases impossible to disable, on school devices? How much has Google spent promoting AI in schools? Has its classroom AI ever encouraged harmful behavior or missed signs that a child was thinking about suicide?

Google's public answer so far is a statement it gave the Journal earlier. It says teachers use its technology "because it solves real classroom challenges and helps improve learning outcomes," that it builds products with educators, and that teachers and administrators decide how the tools get used. Google has also said it built protections for users under 18, including guardrails meant to stop Gemini from acting like a companion and a one-tap link to crisis hotlines when a conversation suggests risk.

## Two switches, and families hold neither

That claim about who's in charge is where the story gets practical, because the controls depend on which AI you mean.

The first is Gemini itself. In August, Google opened a Gemini tab inside Google Classroom to students of every age, after limiting it to students 18 and older. Education Week reported that the tab only appeared in districts that had already allowed students onto the Gemini app, and that districts can switch it off. One Ohio principal said he learned about the change from a news story. Google's own admin guide shows the on-off switch sits with district administrators, and that it covers the Gemini app alone, leaving the AI built into other Google tools on its own settings.

The second is the AI answer at the top of a Google search. Common Sense Media, the nonprofit that rates media for families, says nobody can turn off AI Overviews or AI Mode on a student account: the student, the parent or the school. The senators add that Chromebooks don't come with another browser. Google's own help page does describe one way around the AI answer: after a search, choosing the Web filter returns plain text links. It works one search at a time, it isn't a setting a parent or a school can lock on, and nothing stops a student from going back to the AI answer on the next question.

Common Sense has tested both. It rated Gemini's K-12 app "high risk," partly because it treats a sixth grader and a high school senior alike and will write a full essay when asked. It rated Google's AI search features "unacceptable risk," its worst grade. In its testing, those features completed all 180 homework assignments it tried and failed every one of five severe-harm categories, including one for suicide risk.

Google's competitors drew the line elsewhere. Education Week found that Microsoft keeps its Copilot assistant off for K-12 student accounts unless a school turns it on for students 13 and up, and that Anthropic bars minors from Claude. That left Gemini as the one major chatbot available to students of all ages in schools.

## Crutch or tutor

So what does an always-there AI do to learning? The best evidence comes from a randomized trial in Turkey, published last year in the Proceedings of the National Academy of Sciences. Researchers gave nearly 1,000 high school math students one of two AI helpers for practice sessions. One was a standard ChatGPT-style chatbot. The other was a tutor built with teachers' input that gave hints in place of answers.

During practice, both groups did far better than students with no AI: 48% better with the plain chatbot and 127% better with the tutor. Then the AI was taken away for the exam. Students who had practiced with the plain chatbot scored 17% worse than students who never had it. Students who had practiced with the tutor scored about the same as the no-AI group. The authors' reading is that the plain chatbot became a crutch.

That's one study, in one subject and one country, and the tutor in it didn't beat the no-AI group on the exam. Still, it's a cause-and-effect result, which most of this debate lacks. It points at design, and the AI search answer on a Chromebook works like the plain chatbot.

The exposure is close to universal. In a Common Sense survey this spring of 1,204 U.S. kids aged 9 to 17, 86% said they use or interact with AI, and 75% said they'd used AI summaries in search results. About one in four used AI daily.

## Districts are choosing between a ban and a narrow door

Some school systems have stopped waiting for Google. New York City blocked AI use for about 600,000 students from pre-K through eighth grade this fall, and Los Angeles put a moratorium on AI use on school devices for the year, Axios reported. Clover School District in South Carolina took a middle path this week: it reportedly blocked AI tools on student devices except for one licensed tutoring platform while it sets rules.

Bans have a cost. Axios noted that some educators warn broad bans can pull useful tools from students with disabilities and from kids still learning English. A narrow door keeps a tool like the Turkish tutor and closes off the answer machine.

Take one family. Picture a parent with a seventh grader whose district has the Gemini tab turned on. Before changing anything, she'd want three facts from the school: which AI features are on for her child's account, whether Gemini's Guided Learning mode, which gives hints, is the default, and who at the district can change the settings. Then she'd run a small test at home for two weeks. Her son does one worksheet a week with AI, using only the hint-giving mode, and one without it. The baseline is his last two quiz scores. A good result means his quiz scores hold or rise, and he can explain an answer out loud without the screen. The added work is hers, ten minutes a night checking his explanations. It can fail in three ordinary ways. He switches out of the tutoring mode, which Common Sense found easy to do. He searches the question and reads the AI answer instead. Or he stays in the hint-giving mode and simply asks it to write the thing, which Common Sense found it will do. The mode is a guardrail, not a lock. If his quizzes slip, she can stop the experiment at home tonight. Changing the school's settings takes a district administrator, and the AI search answer has no switch at all.

## What You Can Do With This

### Parents

Ask your school, in writing, which AI features are turned on for your child's account and who controls them. Sit with your kid for one homework session and watch where the answers come from. If your child seems to be leaning on a chatbot for emotional support, talk with them and their school counselor; in the U.S., you can call or text 988 any time for crisis support.

### Students

Try explaining tonight's answer out loud with the laptop closed. If you can't, the AI did the learning for you. Use hint modes when you need help, and treat the AI box at the top of search results as a shortcut that skips the practice. When you do search, choose the Web filter to get plain links for that search, and know it resets the next time.

### Teachers

Give one low-stakes check each week that happens away from the screen, so you can see what students know. Tell families which AI features your class uses and why.

### School board members and district tech leads

Audit your Google Admin settings before Nov. 2, and put the results in front of the board. Ask Google directly how to limit AI search features on student accounts, and get the answer on the record.

## The Bigger Picture

For more than a decade, schools have bought Chromebooks because they're cheap and easy to manage. The trade was that a big company set many of the defaults on the device. That was a small concession when the default was a browser and a word processor. It's a bigger one when the default is an AI that will finish the homework.

My read is that Google's settings will settle this, whatever it writes back. If Google lets districts and parents turn off AI answers on student accounts, or makes hint-giving the default for minors, schools get a real choice between the crutch and the tutor. If it doesn't, more districts will take the blunt option and ban AI outright, including the kinds that help. Google's answer is due Nov. 2.

## References

[U.S. Senate (Blumenthal and Hawley): Letter to Sundar Pichai on Google A.I. tools and students, Oct. 21, 2026](https://www.blumenthal.senate.gov/imo/media/doc/letter_to_google_re_ai_tools_impacting_studentspdf.pdf)
[Sen. Richard Blumenthal: Blumenthal and Hawley demand answers from Google about aggressive AI push in classrooms, Oct. 21, 2026](https://www.blumenthal.senate.gov/newsroom/press/release/blumenthal-and-hawley-demand-answers-from-google-about-aggressive-ai-push-in-classrooms)
[Quartz (Cris Tolomia): Bipartisan senators are pressing Google for answers on AI tools in classrooms, Oct. 21, 2026](https://qz.com/bipartisan-senators-google-ai-schools-letter-102126)
[Politico: Senators press Google on AI in classrooms, Oct. 21, 2026](https://www.politico.com/live-updates/2026/10/21/congress/senators-press-google-on-ai-in-classrooms-01129484)
[Futurism (Joe Wilkins): The tech industry's aggressive push to integrate AI into schools, Oct. 3, 2026](https://futurism.com/future-society/tech-industry-aggressive-google-ai-schools-disaster)
[Education Week (Grace Jiang): Schools caught flat-footed after Google makes Gemini chatbot available to all students, Sept. 1, 2026](https://www.edweek.org/technology/schools-caught-flat-footed-after-google-makes-gemini-chatbot-available-to-all-students/2026/09)
[Common Sense Media (Robbie Torney): What to know about Google turning on Gemini for schools, Aug. 17, 2026](https://institute.commonsensemedia.org/what-to-know-about-google-turning-on-gemini-for-schools)
[Common Sense Media Institute: Gemini K-12 risk assessment, updated Jan. 9, 2026](https://institute.commonsensemedia.org/risk-assessments/gemini-k-12)
[Common Sense Media: AI Use by Tweens and Teens, 2026 census report, March 2026](https://www.commonsensemedia.org/sites/default/files/research/report/2026-ai-use-by-tweens-and-teens-1.pdf)
[Google Workspace Admin Help: Turn the Gemini app on or off](https://knowledge.workspace.google.com/admin/generative-ai/gemini-app/turn-the-gemini-app-on-or-off)
[Google Search Help: AI Overviews in Search](https://support.google.com/websearch/answer/14901683?hl=en)
[Google (Megan Jones Bell and Laurie Richardson): Mental health updates for Gemini, April 7, 2026](https://blog.google/innovation-and-ai/technology/health/mental-health-updates/)
[PNAS via RePEc (Bastani et al.): Generative AI without guardrails can harm learning: Evidence from high school mathematics, July 2025](https://ideas.repec.org/a/nas/journl/v122y2025pe2422633122.html)
[Knowledge at Wharton: Without guardrails, generative AI can harm education, Aug. 27, 2024](https://knowledge.wharton.upenn.edu/article/without-guardrails-generative-ai-can-harm-education)
[Axios (Josephine Walker): AI backlash comes to the classroom, Sept. 7, 2026](https://axios.com/2026/09/07/ai-schools-backlash-bans-new-york-los-angeles)
[Net Gain Edtech (Dallas Kachan): Blumenthal, Hawley demand Google answer for school AI push, Oct. 21, 2026](https://netgain.news/edtech/blumenthal-hawley-demand-google-answer-for-school-ai-push/)