On a humid Monday afternoon in early August 2025, fourteen Korean publishing executives stood in a single-file line outside the National Assembly in Yeouido. Each held a handmade protest sign. They took turns standing alone, an hour at a time, in compliance with Korean solo-protest rules. Inside the building, lawmakers were voting on a bill that would, within hours, erase their core business line.
The bill passed. By nightfall, the Korea AI textbook 2026 program had been stripped of its legal status as official teaching material. Once championed as the most ambitious EdTech reform in Asia, it was now reclassified overnight. Roughly $850 million in government spending had vanished into a four-month classroom experiment. Furthermore, another $580 million in private publisher investment had been funneled into the same program. The new ruling party now classified all of it as “supplementary materials.”
For foreign investors watching Asian EdTech, this was not a minor policy adjustment. Instead, it was a textbook case — pun intended — of how political risk can destroy a billion-dollar industry overnight. Furthermore, it offered an early warning for Meta, Google, and OpenAI, all of which have launched aggressive AI school initiatives in Western markets. In particular, Korea got there first. Then it crashed.
This is the story of how it happened, who lost what, and what the Korea AI textbook 2026 collapse means for the global EdTech industry now scrambling to read the warning signs.
To understand the Korea AI textbook 2026 story, the timeline matters more than the technology. In June 2023, then-President Yoon Suk Yeol’s administration unveiled the AI Digital Textbook Promotion Plan. The vision was sweeping. By 2028, every core subject in Korean elementary, middle, and high schools would be delivered through AI-powered tablets. As a result, students would receive personalized lessons calibrated to their individual learning pace.
Furthermore, the policy carried serious geopolitical framing. Education Minister Lee Ju-Ho promoted it at the UNESCO International Forum on the Futures of Education in late 2024. Korea, he argued, would become the world’s first nation to deploy AI textbooks at national scale. In addition, it would close the gap between rural and urban students. It would even, supporters claimed, reduce dependency on Korea’s notoriously expensive private tutoring institutes — known locally as hagwons.
The numbers behind the bet were enormous. According to government records and reporting by Rest of World, Seoul committed more than 1.2 trillion won — roughly $850 million — to the program. That figure included tablet procurement, teacher training, and the textbook approval review process. Meanwhile, twelve major publishing houses — Cheonjae Education, Visang Education, Dong-A Publishing, YBM, Woongjin Thinkbig, and others — collectively invested an estimated 800 billion won ($580 million) in development.
In September 2024, the Education Ministry approved 76 AI textbook titles. Six months later, in March 2025, the books arrived in classrooms. The Korea AI textbook 2026 era had begun. However, it would last only one semester.
The publishers who joined the program were not small experimental ventures. Rather, they were Korea’s biggest names in K-12 education content. For example, Cheonjae Education is one of the country’s largest textbook companies, with decades of curriculum experience. Similarly, Visang Education had built itself as a tech-forward EdTech publisher since 2004. In addition, YBM dominated English-language education materials. Dong-A Publishing held strong positions in mathematics and science.
Each publisher hired engineers, AI researchers, and product designers to compete. Specifically, the development cycle ran intensely from 2023 to early 2025. As one publisher told Korean media, entire departments were spun up to support the AI textbook business unit. Moreover, executives assumed that government adoption would be mandatory and nationwide. That assumption — as it turned out — was the fatal miscalculation.
For broader context on Korean tech and startup ecosystem dynamics, see Seoulz’s analysis of Korea AI startups and the K-Startup AI League prize program. In particular, Korea’s AI ecosystem was scaling fast across multiple verticals during exactly the period when the Korea AI textbook 2026 collapse unfolded.
When the textbooks launched in March 2025, the first semester became a public relations catastrophe. Students like Ko Ho-dam, a high school junior on Jeju Island, told reporters that classes were repeatedly delayed by login issues and software glitches. Teachers reported factual errors in supposedly AI-vetted content. Furthermore, many noted that personalized recommendations failed to actually adapt to student levels.
In particular, three issues dominated the backlash:
Technical instability. Across the country, classes lost instructional time to login failures, frozen tablets, and incompatibility between school networks and publisher servers. As a result, even teachers who had supported the program initially found themselves spending more time troubleshooting than teaching.
Privacy alarms. AI textbooks collected detailed real-time data on student behavior — including click patterns, hesitation times, and incorrect answers. Parents quickly raised concerns about how this data was stored, who could access it, and whether it complied with Korea’s Personal Information Protection Act. Consequently, parental trust collapsed within weeks of launch.
Teacher overload. A December 2024 survey by the Korean Teachers and Education Workers Union — covering 2,626 teachers — revealed that 98.5% considered existing AI textbook training insufficient. In other words, the rollout had effectively skipped the people who would deliver the program.
The most damning data came from regional adoption maps. In Daegu, a politically conservative city where Yoon had won 75.34% of the vote, AI textbook adoption hit 98% in some subjects. By contrast, in liberal Sejong, only 8% of schools used the textbooks. Furthermore, South Jeolla and Gwangju — both progressive strongholds — sat at 9% and 12% respectively. The Korea AI textbook 2026 rollout had become a partisan flashpoint before it had a chance to prove itself.
The Korea AI textbook 2026 story cannot be separated from the political turmoil of late 2024 and early 2025. In December 2024, President Yoon Suk Yeol declared martial law in a stunning move that triggered impeachment proceedings. By April 2025, the Constitutional Court formally removed him from office. Meanwhile, opposition leader Lee Jae-myung, who had explicitly campaigned on rescinding the AI textbook policy, won the subsequent presidential election.
For Korea’s EdTech publishers, the political timing could not have been worse. Specifically, their entire business model depended on continuity from the previous administration. However, the new ruling Democratic Party of Korea moved decisively. On August 4, 2025, the National Assembly passed a critical amendment. The new law narrowed the legal definition of textbooks to “printed books and e-books.” Specifically, it excluded “learning support software using intelligent information technology.”
According to reporting by The Korea Herald, the new classification took effect immediately upon promulgation. As a result, schools lost the financial framework that had previously funded AI textbook subscriptions. A high school computer science teacher quoted in the report summarized the mood bluntly. Without legal status, the textbooks would not receive funding. As a result, use in class became “almost impossible.”
For investors who had viewed Korea as a leading-edge EdTech laboratory, the speed of the reversal was startling. In particular, less than 18 months had passed between full government endorsement and full government withdrawal. Moreover, the same political volatility that produced this whiplash now hangs over Korea’s broader AI startup ecosystem and any sector that depends heavily on national policy continuity.
Numbers tell the cleanest version of the Korea AI textbook 2026 collapse. After the policy shift to voluntary adoption in early 2025, the program rolled out to roughly 37% of eligible schools during the first semester ending in July 2025. Furthermore, even that figure was concentrated in conservative regions where political support remained strong.
However, the August legal status change accelerated the decline dramatically. By the second semester beginning in September 2025, adoption had fallen to 19%. In other words, in less than three months, half of the participating schools had dropped the program. According to data cited by Rest of World, the trajectory continued downward through the autumn. As a result, what began as a flagship national initiative had effectively become a niche optional supplement.
The implications for publishers were brutal. Without legal textbook designation, schools could not access government subscription budgets. Consequently, any publisher revenue had to come from voluntary school spending — a category Korean public schools rarely use for ongoing software costs.
The financial damage to Korea’s textbook publishing industry has been documented in detail by The AI, a specialist Korean tech publication. Specifically, total private development costs across the twelve participating publishers reached approximately 800 billion won — roughly $580 million. The Education Ministry, in October 2025, agreed to refund only 4.88 billion won in approval review fees. In addition, that refund excluded publishers whose first-round textbooks had already been adopted.
Three early casualties emerged within the first year of the collapse:
Furthermore, the legal response began almost immediately. In April 2025, several publishers filed an administrative lawsuit against the Education Ministry, citing losses from low adoption and policy inconsistency. Then, on December 7, 2025, the Korea Textbook Association — together with YBM, Cheonjae Education, and Dong-A Publishing — formed the AIDT Emergency Response Committee. The committee filed a constitutional complaint with the Constitutional Court of Korea, arguing that the legislative reclassification “threatens their survival” as businesses.
Hwang Geun-sik, chair of the Textbook Development Committee, captured the industry’s mood in his comments to Rest of World. Companies that trusted the government and invested saw the market suddenly disappear. As a result, staff cuts have become unavoidable.
For foreign EdTech operators and investors, the Korea AI textbook 2026 saga delivers four uncomfortable lessons.
First, government-anchored EdTech carries existential political risk. The binary classification of educational materials — official textbook or not — amplifies that risk dramatically. In particular, when policy shifts, an entire revenue line can disappear in a single legislative session. Meanwhile, in the United States, Meta, Google, and OpenAI are all building school-facing AI products that rely on district-level or state-level adoption. Although the political mechanism differs, the dependency structure does not.
Second, classroom-grade AI is harder to ship than enterprise AI. Korean publishers had access to substantial AI research talent and government coordination. However, the textbooks still launched with factual errors, latency issues, and adaptive systems that did not meaningfully personalize. As a result, the Korea AI textbook 2026 collapse highlights how unforgiving classroom deployment can be compared to corporate or consumer use cases.
Third, parental trust is the gating factor. A petition from the parent group “Political Mamas” gathered 56,505 signatures opposing AI textbooks in May and June 2024. Furthermore, parents successfully reframed the debate from “innovation” to “screen time and privacy.” For instance, every Western EdTech company building K-12 AI tools should expect a similar reframing. In particular, the parental constituency moves faster than the regulatory one.
Fourth, teacher buy-in is non-negotiable. When 98.5% of surveyed teachers said training was insufficient, the policy was already failing. Teachers are the deployment infrastructure. As a result, EdTech firms that bypass them — or assume top-down mandates will overcome resistance — repeat Korea’s mistake.
For broader analysis on Korea’s AI investment landscape and where capital is now flowing, see Seoulz’s coverage of Korea AI chip startups and Upstage, Korea’s first generative AI unicorn. In particular, capital that might once have flowed to government-backed EdTech is increasingly redirecting toward enterprise and consumer AI plays where political dependency is lower.
Despite the collapse, the Korea AI textbook 2026 story is not entirely a postmortem. Specifically, several publishers are pivoting toward private hagwon partnerships, where adoption decisions sit with individual academy operators rather than government ministries. Meanwhile, others are exporting their AI textbook engines to Southeast Asian markets where policy environments are friendlier.
Furthermore, the underlying technology — adaptive learning algorithms, multilingual content engines, real-time feedback loops — has not disappeared. Instead, it is being repackaged for B2C consumer subscription apps. For example, Korean EdTech startups like Qanda (Mathpresso) and Riiid have continued to grow through this period, partly because they never depended on government textbook designation in the first place.
In addition, the AI Framework Act may eventually create a clearer regulatory environment for AI in education. Korea’s comprehensive AI governance law is expected to take full effect in 2026. As covered in Seoulz’s analysis of Korea’s AI commercialization push, Seoul is still spending heavily on applied AI. However, the deployment model is shifting away from compulsory mandates. Instead, voluntary, market-driven adoption is now the preferred approach.
For foreign EdTech investors and operators evaluating Korea, the practical takeaways from the Korea AI textbook 2026 collapse are straightforward.
The K-12 government channel is now effectively closed for AI-driven materials. As a result, betting on national mandates carries unacceptable political risk. However, the private tutoring market — Korea’s hagwon ecosystem, valued at roughly $22 billion annually — remains open. In particular, hagwons make purchasing decisions independently and have shown strong willingness to adopt AI tools that demonstrably improve student performance on standardized tests.
Meanwhile, the consumer subscription channel continues to expand. Specifically, Korean parents who lost trust in school-deployed AI have not lost trust in AI tutoring per se. Instead, they are paying directly for apps with transparent pricing and parental controls. As a result, the addressable market for B2C EdTech may actually be larger now than before the AI textbook program launched.
Finally, the Korea AI textbook 2026 episode is a useful pricing input for any Western EdTech firm considering school-district partnerships in the United States or Europe. Specifically, the political risk premium needs to be modeled explicitly. In addition, multi-year contracts with sovereign or sub-sovereign buyers should include clear termination terms and IP recovery mechanisms.
The Yeouido protests outside the National Assembly may have ended quietly. However, the lessons of those protests — about political risk, parental power, teacher voice, and the brutal economics of mandate-dependent business models — will travel far beyond Korea. In particular, every EdTech founder pitching “the future of learning” in 2026 should have a clear answer to one question: what happens when the policy that funds you disappears?
For Korea, that question is no longer hypothetical. For everyone else, it is just a matter of time.
On a Wednesday afternoon in February 2026, Costco shoppers in suburban California walked past a…
On a cold Tuesday morning in late December 2025, a press release dropped from College…
Article Body The Korea wedding industry collapse has a face. It is a Tuesday afternoon…
The Asan Factory Apple Cannot Replace The Korea OLED display 2026 story begins with a…
On a Wednesday morning in late April 2026, South Korea's Ministry of Health and Welfare…
On a Wednesday at noon in November 2025, hundreds of people lined up along Bishop…