Photo: Allen Y / Unsplash
Defaults are the most powerful design decision in software, and bundles are the most powerful business decision. Earlier this month Google combined the two in a way that set off a fierce reaction. According to The Associated Press, the company said students using its software at school will lose access to Google search unless their school system also gives them access to Google’s AI tools. Google framed the change as handing control to schools. Critics, including a former head of the Federal Trade Commission, called it something closer to a squeeze. For anyone building AI products, especially for kids, families or institutions, this is a case study worth reading closely.
What Google changed
Here is the sequence. In August, Google turned on its Gemini chatbot for many K-12 students using Google Classroom. Before that, it had allowed Gemini only for users 18 and older. The move drew pushback from educators, child development experts and parents.
In its new policy, announced in a blog post by Chris Phillips, Google’s vice president and general manager of education, AI will no longer be switched on by default for school administrators. Gemini, Gemini Notebook, Gemini in Classroom and Google search will all be off for K-12 student accounts until administrators turn them on. The catch is that the two are linked: opting out of the AI tools also means opting out of Google search. Phillips wrote, “We believe schools should have control over AI in their classrooms.”
Google is the dominant education technology provider in U.S. schools. It says 150 million students and teachers worldwide use Google Classroom, and many students work on school-issued Chromebooks.
The backlash
The response was swift. Lina Khan, the former chair of the Federal Trade Commission, wrote on social media that Google was using its dominance in search to pressure schools into adopting its AI. Josh Golin, executive director of the children’s advocacy group Fairplay, said the attempt to push schools into its AI products was likely to backfire, and argued that the company’s school technology is aimed at building lifelong brand loyalty more than learning. He urged schools to turn off Google’s AI tools as soon as possible and suggested they consider switching off Google search as well in favor of more privacy-protective alternatives. Google did not respond to AP’s requests for comment.
The context matters. New York City and Los Angeles, the country’s two largest school districts, have temporarily banned generative AI tools in elementary and middle schools while they assess safety and write policies. Many students already use AI for schoolwork, it makes cheating easier, and researchers are still studying how it affects children’s developing brains.
Why this matters beyond Google
Most founders will never have Google’s leverage. That is exactly why this episode is useful. It shows, at full scale, the three forces every AI product now has to manage.
- Defaults signal values. Turning AI on for children by default in August, then off by default now, tells customers who the product was really designed around. Users notice which way the switch starts.
- Bundling can feel like coercion. Packaging a new, controversial feature with an essential, trusted one may lift adoption numbers. It can also convert a product debate into a fairness debate, and invite regulators into the conversation.
- The buyer is not the user. In schools, administrators buy, teachers deploy, students use and parents worry. A decision that pleases one group can alienate the other three.
The opening for startups
When an incumbent loses trust, it creates room. Golin’s advice to schools explicitly pointed toward more privacy-protective alternatives. That is a market signal. Districts that pause or switch off AI tools still need search, research and writing support for students. Teachers still need help with planning and feedback. Parents still want visibility into what their children are using.
Founders who can offer clear controls, honest data practices and tools built around learning rather than engagement have a real opening. But the bar is high, and it should be. Schools that just watched a giant change its rules twice in three months will ask hard questions of anyone new.
If I were building in this space, I would start with teachers rather than with features. Sit in on classes. Ask what a good week looks like and what a bad one looks like. Learn which tasks eat their evenings and which tools they already distrust. Then build the smallest thing that gives them time back without giving students a shortcut around the thinking. A product that makes a teacher’s job easier and a student’s learning deeper is a product a district can defend at a school board meeting. That is the real sales cycle in education: not the demo, but the moment an administrator has to explain your tool to a room full of parents. Design for that moment from day one, and you will not need a bundle to win.
A trust-first checklist for AI product decisions
- Ship sensitive features off by default. Especially for minors, health, finance or anything high stakes. Let customers opt in with full information.
- Unbundle the controversial from the essential. Let customers adopt your new AI feature on its own merits. If it is good, it will win without holding something else hostage.
- Explain the switch in plain language. Tell administrators exactly what turns on, what data it uses and what changes for users.
- Design for every stakeholder. Map the buyer, the deployer, the end user and the people responsible for that user. Test each decision against all four.
- Publish your data practices. What you collect, how long you keep it, whether it trains models and how a customer can delete it.
- Do not change the rules quietly. Announce policy changes early, give customers time to prepare and keep a public changelog.
- Get legal advice on bundling. If your product is a leader in its category, tying products together can raise competition concerns. This is general information, not legal advice; talk to a qualified attorney.
Questions school buyers will ask you
- Is your AI feature on or off when we first set up student accounts?
- Can we use your core product without the AI features?
- Does student data train your models?
- How do you support teachers in preventing AI-enabled cheating?
- What will you do if our district, or our state, restricts generative AI next semester?
If you cannot answer all five in a sentence each, you are not ready to sell into schools.
What to watch next
- District decisions. How many school systems switch off Google’s AI tools, and whether any give up Google search to do it.
- Regulatory interest. Whether the criticism from a former FTC chair turns into formal scrutiny of the bundling.
- Big-district policies. What New York City and Los Angeles decide once their reviews of generative AI in younger grades are complete.
- Research on kids and AI. New findings on learning and brain development will shape what parents and schools will accept.
- Competitors. Whether rivals use this moment to court schools with unbundled, privacy-first offerings.
The bottom line
Every AI company is now making the same choice Google faced: grow adoption by default, or earn it by consent. The first is faster. The second lasts. In a week when many of us are thinking about the teachers who shaped us, I would encourage every founder building for classrooms to remember who the product is ultimately for. Build the version a teacher would be proud to recommend, and the adoption will come.
Source: The Associated Press
Quick check
How much did you catch? Tap a question to see the answer.
1. Under Google’s new policy, what happens if a school opts out of Google’s AI tools for K-12 students?
Answer: Students also lose access to Google search.
2. How many students and teachers worldwide does Google say use Google Classroom?
Answer: 150 million.
3. Which two school districts have temporarily banned generative AI tools in elementary and middle schools?
Answer: New York City and Los Angeles.
