Over Labor Day weekend, Springfield Public Schools, a district in western Massachusetts that serves twenty-three thousand students across some sixty schools, fell prey to a cyberattack that shuttered classes for four days. Staff members lost access to students’ contact information, transcripts, and medical records; internal phone lines, security-camera systems, and even copy machines went down. Teachers and students couldn’t use their school-issued laptops or log into their e-mail or other school applications from home devices. Springfield’s schools use the Microsoft 365 Education and PowerSchool software platforms to house most of the stuff of everyday learning—schedules, assignments, slide decks, readings, handouts, quizzes—but everything stored there was now out of reach. In the midst of the crisis, the district circulated an F.A.Q. with a rather exotic recommendation: “Students and teachers may use books, notebooks and paper while technology teaching tools are restored.”

Meg Magrath-Smith is an instructional leader for English language arts and social studies at one of Springfield’s schools. On the second day of the outage, she sent some colleagues a text message warning that, whenever they did return to their classrooms, they’d likely “be teaching like it’s 1999,” she told me. “Except not, because in 1999 we had copy machines and transparency machines” (overhead projectors). The teachers and students of the twentieth century also had textbooks. Magrath-Smith’s school, like many in the United States, does not.

The cyber-invasion of Springfield is part of a years-long, escalating pattern of ransomware attacks on K-12 school systems, which are rich in potentially valuable student data and relatively poor in cybersecurity resources. But the emergency that the attack created also reflects how public schools in the U.S. have effectively surrendered much of their pedagogy, curricula, and operating systems to a handful of trillion-dollar technology companies. Foremost among these is Google, whose Chromebook laptops and Google Workspace for Education software have penetrated a super-majority of districts; Apple and Microsoft’s devices and classroom software hold a more modest share of the K-12 market. “Google and Apple and Microsoft have become the scaffolding on which many U.S. schools function,” Natasha Singer writes in her new book, “Coding Kids: Big Tech’s Battle to Remake Public Schools.” “The tech industry’s growing role in the classroom has become too big and too integral, it seems, to interrogate.”

Singer is a tech reporter for the Times, in which she publishes essential coverage of Silicon Valley’s ongoing colonization of the K-12 classroom. “Coding Kids” is an even-tempered yet withering narrative of how the nation’s public schools came under the spell of private corporations, exemplifying what the social psychologist Shoshana Zuboff defined as surveillance capitalism. Google has repeatedly been accused of illicit collection of students’ data; back in 2014, the company acknowledged that it had scanned students’ e-mails in its education platform for targeted-advertising purposes. Most recently, Google settled a class-action lawsuit alleging that its education platform improperly captured and stored the biometric data—the voices and faces—of students in Illinois. (The company did not admit wrongdoing.) Currently, Microsoft faces a proposed class-action lawsuit in Washington State alleging that the company “collects, extracts, and generates vast troves of student data, converts that data into hyper-detailed psychographic profiles of students, and uses that data for its own commercial and other purposes, enriching itself at students’ expense.” (Microsoft has previously stated that the data is used solely for educational purposes.) One might say that the plaintiffs in these cases, too, saw themselves as victims of a cyberattack.

It’s astonishing, given these two companies’ track records, that a large majority of K-12 students spend many or most of their school hours staring at Microsoft and Google’s hardware and software, which are now loaded with generative-A.I. tools that can contribute to cognitive atrophy and loosen our collective hold on reality. Research strongly indicates that such a state of affairs is in conflict with students’ right to privacy and to a free education, and actively threatens their intellectual and social-emotional development. Yet the situation persists, apparently because it aligns with the predilections of wealthy technologists—who prefer, for example, that “books, notebooks and paper,” those unhackable learning platforms that somehow sufficed for centuries, should become nearly obsolete, the teaching tools of last resort.

“Coding Kids” begins in 2013—the year of Vine, Edward Snowden, and the Xbox One—with the founding of a new educational nonprofit, Code.org, to promote the teaching of computer programming in K-12 schools. Funded primarily by Microsoft and Google, Code.org launched with a video titled “What Most Schools Don’t Teach,” featuring an unblinking Mark Zuckerberg, a co-founder of Facebook and, later, of Meta; Jack Dorsey, a co-founder of Twitter, back when he looked less like the leader of a breatharian yoga cult and more like a junior sales rep at Siebel Systems; and, for some reason, Will.i.am, of the Black Eyed Peas, who was credited as “Now Taking Coding Classes.” These V.I.P.s were assembled to recruit a new generation of computer programmers from the nation’s K-12 schools. “1 million of the best jobs in America may go unfilled,” the video warned, “because only 1 in 4 schools teach computer science.”

Also providing a cameo was Bill Gates, the centi-billionaire co-founder and former C.E.O. of Microsoft, and the chair of the Gates Foundation. Gates had already spent years trying to impose a highly bureaucratized, privatized, and data-fixated vision of teaching and learning in K-12 schools. Starting in 2009, for example, the Gates Foundation poured two hundred and ninety million dollars into a multiyear program to improve teacher effectiveness using metrics that relied heavily on standardized-test scores. (An encyclopedic RAND Corporation analysis of the program, published in 2018, found that it did little to improve teacher or student performance, while adding significant administrative burdens to participating schools.) The Gates Foundation was also a lead funder of the hundred-million-dollar nonprofit inBloom, né the Shared Learning Collaborative, which emerged, in the early twenty-tens, with the aim of building a centralized database of highly sensitive student data: Social Security numbers, disciplinary records, parents’ marital status, and more. The data was intended to be stored on Amazon’s cloud servers and made available to third-party venders for the creation of customized ed-tech products. In 2014, though, after widespread parental backlash, inBloom shut down.

It’s surprising that the inBloom debacle doesn’t appear prominently in “Coding Kids,” which arguably devotes too much of its real estate to the starry, fake-it-till-you-make-it likes of Code.org. “What Most Schools Don’t Teach” eventually logged more than sixteen million views and drove heavy traffic to Code.org’s website, even though, as Singer writes, Code.org did not yet “have its own programming lessons, or any other tangible education product or service, to offer its eager audience of teachers and students.” By the end of 2013, Code.org had managed to launch an annual event, the Hour of Code, which offered free “tutorials”—drag-and-drop puzzles featuring, say, Disney princesses or Minecraft voxels or Barbie—in classrooms and in Apple and Microsoft retail stores. The organization later worked with the College Board to develop “instructional resources” for a new Advanced Placement class, Computer Science Principles, which launched in 2016. And millions of public-school teachers and students have used Code.org’s videos and worksheets.

Educators and parents who oppose the encroachment of Big Tech into schools are sometimes criticized for lacking a vision of the future that isn’t either dystopian or Luddite-lite. Yet the techno-vangelists of Silicon Valley aren’t necessarily gifted with foresight, either. The objective of the “Learn to code” messaging blitz of the twenty-tens was to equip children with the skills they would need for the tech-industry jobs of tomorrow, but this goal has been significantly thwarted by the tech industry itself, in which generative and agentic A.I. has streamlined much of the work of writing, testing, and correcting code. Kids who might have caught the programming bug during an Hour of Code in the twenty-tens are now entering a dismal market for recent computer-science graduates, as A.I. has made many entry-level tech jobs obsolete. (Tech-sector layoffs and federal-workforce cuts have also contributed to the dismal employment picture.)

This trajectory is embodied by Code.org’s most famous champion, Barack Obama. In 2013, Hadi Partovi, a co-founder of Code.org, persuaded the President to record a video to kick off Computer Science Education Week and the first Hour of Code. “Learning these skills isn’t just important for your future—it’s important for your country’s future,” Obama said in the video. The following year, Obama visited a high school in Newark to participate in an Hour of Code, where he donned a Code.org baseball cap and used a laptop decorated with a Code.org sticker. By 2025, however, Obama was opining that “unless you are really good—like, one of the top one per cent in terms of understanding how to code—you’re better off with a liberal-arts education.” Partovi, who is an alumnus of Microsoft, has remarked that A.I. has obviated much of the “busywork of computer science.” In June, his organization announced that it was “entering its next chapter,” and rebranded itself as CodeAI.

The nonprofit’s website currently features a list of top donors: Microsoft, Google, Amazon, the private-equity firm Vista, the Gates Foundation, and the Ballmer Group, the last of these named for another former Microsoft C.E.O. and centi-billionaire, Steve Ballmer. Each of these oligarchic fun houses has staked a significant share of its future wealth—and, together, the future of the global economy—on the rapid expansion of A.I. technology.

The coding missionaries of the twenty-tens made errors both practical and philosophical. They were wrong, obviously, about computer-programming skills as a guarantor of students’ future employment prospects. But that miscalculation was, itself, premised on the unexamined assumption that a K-12 public education should be essentially vocational and therefore—almost by definition—aligned with corporate interests. In 2021, David Banks, then the incoming chancellor of New York City Public Schools, gave a speech in which he lamented that the city’s schools were producing graduates who “don’t even have the skills to go into industry.” Teachers, he went on, were failing to address an urgent question: “What is it like to work at a place like a Google or a Microsoft?”

But teachers and students do, in their way, work for Google and Microsoft. They supply the companies with valuable data and test-drive their products. More than any other entity, and perhaps to Bill Gates’s great despair, it is Google that determines the conceptual and material framework of K-12 public education in the United States. In many respects, Google Workspace for Education—where students’ profile photos, passwords, grades, attendance records, e-mails, chat logs, and more can be stored—is inBloom by any other name. Google’s control over hardware, software, and knowledge-gathering in public schools creates conditions under which students are compelled, as two Swedish researchers wrote a decade ago, “to participate in the commodification of their digital labour and algorithmic identities.” Teachers, parents, and students with ethical or privacy-based objections to Google’s educational infrastructure have little in the way of other options. In many districts, opting out of Google Classroom would be close to opting out of public school altogether.

“Educators and policymakers are no longer making decisions about technology in the classroom, though they may feel like they are,” Emily Tavoulareas, an independent researcher who studies technology and public policy, stated in a recent address to the Fairfax County School Board, in Virginia. “Companies are making those decisions, because digital technology has become invisible infrastructure that our schools now depend on.”

A crystallizing example of this dynamic came earlier this year, when Google added an all-ages version of Gemini, its suite of A.I. tools, to school-issued Chromebooks. In some districts, the change took even top-level administrators by surprise, and it may have splashed accelerant on an already intensifying blaze of anti-A.I. sentiment. A large survey by the Institute for Family Studies, for example, shows that seventy-five per cent of parents of K-12 students support A.I. moratoria. In an NBC News survey, fifty-three per cent of respondents said that A.I. will do “more harm than good” in K-12 education. Even Gates himself, who just last year was musing that A.I tutors might replace most teachers, recently expressed concerns about A.I. eroding young people’s emotional well-being and critical-thinking skills.

In theory, opting out of A.I.—which its own architects constantly warn might kill us all—should be entirely feasible. In most public schools, eschewing A.I. should be tantamount not to teaching like it’s 1999 but to teaching like it’s a few years ago. In the summer, the Fairfax board passed motions to limit the use of laptops, A.I., and other ed-tech products in its elementary and middle schools, and, on September 2nd, the nation’s two largest public-school systems announced major restrictions on the use of generative-A.I. products in their classrooms. The Los Angeles Unified School District now blocks stand-alone A.I. tools across all grades. (In February, the F.B.I. raided the home of the district’s then superintendent, Alberto Carvalho, a CodeAI board member, as part of an investigation of a defunct A.I. company that was developing a chatbot for students. Carvalho has denied any wrongdoing.) And New York City Public Schools established a one-year moratorium on all generative-A.I. use through eighth grade; Gemini, for one, is now disabled in students’ Chromebooks throughout the city, along with dozens of stand-alone chatbots. Both cities are also pressing new screen-time restrictions. In New York, students through second grade will—officially, at least—have no one-to-one time with devices; students between third and fifth grade will be limited to thirty minutes per day; and middle schoolers will be capped at forty-five minutes.

“The tech industry wants us to believe that A.I. in early education is not only inevitable but that it is necessary,” New York’s mayor, Zohran Mamdani, said during a press conference announcing the moratorium. “We do not see it that way. I have yet to see a study showing that A.I. is beneficial for students in elementary and middle school—with the exception, of course, of research sponsored by the very corporations that stand to profit.”

But, well into the second week of the new school year in New York, questions and concerns persist about perceived loopholes in the new guidance. The one-to-one screen-time limitations do not apply to the bombardment of “assessments” and “diagnostics” to which students are subjected via gamified online platforms such as Acadience and the dreaded i-Ready. Teachers are still generally permitted to use A.I., with a few vague exceptions (including for grading and “behavior monitoring”), so students might still receive A.I.-generated assignments and feedback. This is a blatant double standard, and one guaranteed to breed contempt in A.I.-averse students. And, although Gemini has been disabled on students’ Chromebooks, they are still free to use Google.com, which increasingly resembles a chatbot more than a search engine.

It’s also unclear why, starting in ninth grade, A.I.-inflected ed-tech in New York remains something of a free-for-all—after all, one presumes that Mamdani has yet to see a study showing that A.I. is beneficial for students in high school, either. There are plenty that show it isn’t. A recent large study of Chinese students between the ages of twelve and eighteen found that, after six months of using A.I., they completed their homework assignments more accurately and much faster than before—but that their test scores were twenty per cent lower than those of peers who didn’t use A.I. to do their homework. This finding aligns with virtually everything we know about A.I.’s adverse effects on cognition and working memory.

The city is nonetheless going ahead with a quintet of outside A.I. pilot programs for high schoolers, which may reach as many as fifty thousand students. These pilots include Quill, a language-arts platform that offers students “A.I.-powered feedback” to “revise their responses to strengthen their writing.” There’s Brisk Teaching, whose website features tools for teachers including a “Quiz Maker,” a “Rubric Generator,” and a “Lesson Plan Generator.” And there’s Intel® AI-Ready Schools, a platform for public education that just happens to have a registered trademark in its name. But, if New York City Public Schools is indeed sending mixed messages on A.I., it’s certainly doing so in élite company. Less than two weeks after New York and Los Angeles announced their moratoria, the Gates Foundation announced a one-billion-dollar commitment to “help build and deliver equitable A.I.,” reserving forty per cent for A.I. in education. ♦