CHICAGO. It was raining, the kind of cold November drizzle that thins a crowd, and about 30 people were still outside Google's West Loop office on Wednesday evening chanting "Hey Google, nice try, we don't want your Gemini." One was a 16-year-old junior at Kenwood Academy named Reese Mackey, who told Chalkbeat that students are "owed lessons and feedback from the people CPS hired," not "a bot whose only job is to be a yes man."
They were not there because Google shipped something. They were there because Google took something away, or threatened to. Earlier this month the company told U.S. school districts that if they disable AI features in Google Search, they cannot have Google Search. AI is now so embedded in the product, the argument goes, that there is no non-AI version left to hand over.
The Choice Wasn't a Choice
Google's framing was generous on paper. Schools were getting "more choice, not less," the company said, because administrators could now switch off Search, Gemini, and Gemini Notebook by default for their students. What the statement did not include was the obvious fourth option: Search on, AI off.
That gap is the whole story. A district that wants students to type a query and read ten blue links has exactly one lever left, and it is a string. Add "-ai" to every query and the AI overview disappears. Expect a 12-year-old on a Chromebook to do that 40 times a day, in every class, on every assignment, and you have not designed a control. You have designed a gotcha.
- What Google announced: schools may disable Search, Gemini, and Gemini Notebook by default.
- What Google did not announce: a way to keep Search while disabling AI.
- The only workaround: appending "-ai" to every single query.
- Who decides what "embedded" means: Google, unilaterally, with no published technical spec.
Chicago Already Learned This Lesson in August
This is not CPS's first round with Gemini. The district had planned to put Gemini in front of every high school student districtwide for the 2026-2027 school year, and backed off after community pushback, according to the district's own AI Access Overview. That retreat followed organizing by Illinois Families for Public Schools, a grassroots nonprofit that mobilized parents who had not been consulted before the rollout was announced.
Since then CPS has approved at least three generative AI tools for classroom use, Gemini, Microsoft Copilot, and Gemini Notebook, under contracts barring vendors from training on student data. The district that just blocked student access to generative AI also holds signed contracts for generative AI. Neither fact is a policy.
The district's answer to Google's ultimatum is a study in hedging. A CPS spokesperson said the district is "actively exploring all options to independently suppress" the AI overview at the top of Google results, and that if suppression proves impossible it will "look to other search solutions." That is the sound of a customer realizing the product they bought for fifteen years is no longer the product they are being sold.
Behind the chants is something uglier. Ars Technica reports that a lawsuit filed by a teen describes Gemini as a "suicide coach." Those allegations will be litigated for months, and any honest reading has to hold two things at once: a chatbot is not a therapist, and a company that ships a conversational model to minors at national scale has taken on obligations its terms of service do not discharge.
Google's public answer is Guided Learning, a feature it says it is piloting that walks students through a problem step by step instead of handing over an answer. Reasonable product idea. It is still not a policy, not a parental consent mechanism, and not an answer to what happens when the step-by-step tutor is wrong in front of 300,000 kids on the same morning.
A Moratorium Is a Bug Report, Not a Ban
The core demand is a three-year pause on generative AI in CPS classrooms until the district writes actual policy. Framed by critics as technophobia, it reads more like requirements gathering: three years is roughly how long a large district needs to procure, pilot, and evaluate anything, and the group attached specifics instead of vibes.
- A three-year moratorium on generative AI in classrooms, paired with "human-centered education."
- A chief privacy officer at the district, with authority over student data.
- Screen time limits across all grade levels, not just elementary.
- A ban on biometric collection, including face scans and fingerprints.
About half of the incoming Chicago Board of Education, including president-elect Jessica Biggs, signed that pledge during the campaign. Board member Michilla Blaise, reelected in District 5b, says AI is widening learning gaps on the West Side. New York City, under Mayor Zohran Mamdani, banned student-facing AI from preschool through eighth grade before the school year began, and by late October its Education Department had flagged nearly 40 products, Grammarly and Kahoot among them, for their AI features.
Here is the skeptical read. Google is not lying that AI is inseparable from modern Search, but inseparability is a design decision, not a law of physics, and design decisions have owners. "More choice, not less" survives right up until someone requests the one configuration left out. If Google wants to be in classrooms, it should publish a spec: what can be disabled, at which admin level, verified by whom. Until then districts are not choosing between AI and no AI. They are choosing between Google's AI and somebody else's search box, which is exactly the outcome the company is trying to prevent.
The protests were small, roughly 30 people in Chicago, a parallel rally in Cambridge, Massachusetts, and actions Ars counted in seven cities. Small is not fringe. Nearly 40 jurisdictions passed ballot measures limiting data center expansion in last week's midterms, and that same skepticism has now found the classroom.
Comments