Do you have a right not to use AI? Google's standoff with schools says that, for now, you barely do

THE THESIS. Saying «no» to AI is becoming a technical privilege, not a right. Google's dispute with schools is not about whether AI is good or bad for learning. It is about who decides. When a dominant company fuses an essential service, like web search, with a new product, like AI-generated answers, users lose something valuable: the ability to choose. We are optimistic about AI in education over the long term. That is exactly why we think it should earn its place in the classroom, not be imposed by default.

WHAT ACTUALLY HAPPENED. On November 3, Google announced a change to school accounts for US students aged 5 to 18. Gemini, Gemini Notebook, Gemini in Classroom and Search itself will stay off until a school administrator turns them on. According to The New York Times, Google's education chief, Chris Phillips, said that «Search and AI are deeply integrated». He added that if a school does not enable AI, its students will not get Search. The backlash was immediate. According to CNBC, New York City Mayor Zohran Mamdani accused Google of using «its monopoly power to force AI into our public schools». Senator Bernie Sanders called it «extortion». Former chair of the FTC, the US competition agency, Lina Khan warned about leaving «an essential service» to «an unregulated monopolist». Google later clarified that each service can be enabled separately. A school can turn off Gemini and keep Search. But according to Ars Technica and MediaNama, there is no version of Search without AI Overviews, the automatically generated summaries that appear above the links. The Albany school district in New York confirmed it: it could disable Gemini and AI Mode, but not those summaries. Google itself says its clarification «did not represent a policy change». Senator Ed Markey has asked Alphabet to let schools keep Search without AI, with answers due by November 30.

OUR READING: THE PROBLEM IS BUNDLING. The school case is the most visible version of a pattern we have tracked for months. Meta added its assistant to WhatsApp with an icon that, as the company itself said in 2025, «is not possible to disable». Microsoft raised Microsoft 365 prices in Australia after adding Copilot. Australia's consumer watchdog, the ACCC, has sued the company. It alleges that Microsoft hid from about 2.7 million customers that a «Classic» plan without AI existed at the old price: the Personal plan rose from A$109 to A$159 a year, 45% more. Microsoft says it is reviewing the claim. Google, for its part, maintains that AI is a core part of Search. The mechanism is the same in all three cases. AI does not arrive as a product you choose. It arrives attached to one you can no longer stop using. That is why this is about competition, not nostalgia. If 90% of searches go through Google, then Google making its AI the default is not just a new feature. It is a way to hand out market share in the next big business.

THE PRECEDENTS: WE HAVE SEEN THIS BEFORE. Twenty-five years ago, US courts found that Microsoft had illegally protected its Windows monopoly, in a case centred on how it bundled Internet Explorer. Brussels later imposed a browser choice screen. The more recent parallel is the Android case. In 2018 the European Commission fined Google for forcing phone makers to pre-install its search engine and browser. In July 2026, the EU Court of Justice upheld the €4.125 billion fine with no further appeal. It rejected all six of Google's arguments. The legal lesson is clear: tying a new product to a dominant one can be an abuse of dominance. The Digital Markets Act (DMA, the EU law that sets up-front obligations for big digital «gatekeepers») already applies to AI. In July 2026, the Commission ordered Google to open Android features it had reserved for Gemini to rival assistants such as ChatGPT or Claude. This year Brussels also said, in preliminary findings, that Meta may have broken competition rules by shutting third-party assistants out of WhatsApp. In April the Commission rejected Meta's offer to charge them for access instead. Another front remains open: a DMA case on whether Google favours its own services in Search, including Gemini's role in AI Overviews. And an alliance of independent publishers has complained to Brussels that they cannot keep their content out of those summaries without disappearing from Search. It is the same dilemma schools face.

THE RIGHT THAT (ALMOST) DOES NOT EXIST. Is there a legal right not to use AI? Today, only partly. Article 22 of the General Data Protection Regulation (GDPR) protects people from decisions made solely by a machine that have legal or similarly significant effects, such as refusing a loan. It gives a right to ask for a human to step in. The EU Court of Justice widened its scope in the 2023 SCHUFA case, on credit scores. The EU AI Act adds transparency, such as knowing you are talking to a chatbot. It also gives a right to an explanation when a high-risk system makes a decision about you. But neither law gives a general right to an AI-free product. They protect you from automated decisions that affect you, not from a summary you never asked for. Competition and consumer law are filling the gap. The Australian lawsuit does not say Copilot is illegal. It says hiding the AI-free alternative was misleading. That is the most promising path: not banning AI, but requiring that the alternative exists and is easy to see. The market shows it can be done. In February, Firefox added a switch that turns off all its AI features. DuckDuckGo offers a no-AI version of its search engine. If they can do it, so can Google.

WHO PAYS FOR SAYING NO, AND WHAT IT MEANS. Opting out has a real cost. A school that rejects Google's AI loses Search, or must pay for third-party filters such as Lightspeed that redirect searches to plain web results. Poorer districts cannot always afford that. Some news outlets, such as the Stanford Daily or the WEHCO group, ban generative AI in their content. They protect their credibility that way, though it may cost them productivity against rivals that use it. New York City has set a one-year moratorium on student-facing generative AI up to eighth grade. Our reading is the usual one: measured optimism. In the short term, imposing AI by default carries real risks. Common Sense Media, a child-safety nonprofit, rated Google's AI search features an «unacceptable risk» for minors. And according to Stanford, there is still not enough solid research showing that AI improves learning. In the long term, we believe AI could be the best tutor that has ever existed: patient, personalised and free for millions of children. That promise is exactly why it should not be forced on anyone. Technology that truly adds value does not need to hide its off switch. What we will watch: Google's reply to Markey on November 30, the EU case on AI Overviews, and whether the EU's planned Digital Fairness Act turns the «no-AI option» into an obligation. The question is not whether we will use AI. It is whether we will be able to choose when.