Published July 2026.
Duolingo spent the past year running an experiment in how far a consumer app can be pushed toward AI-driven efficiency before its own users and its own business model push back. In April 2025, the company announced it was going “AI-first.” A year later, that strategy had unwound in two ways: AI usage quietly disappeared as an employee performance metric, and more consequentially for teachers, Duolingo for Schools, its free classroom tool, is now set to shut down entirely on July 31, 2027. Both reversals trace back to the same underlying tension, and together they offer a useful case study for language teachers thinking about how much to lean on AI tools in their own classrooms.
What Actually Happened
On April 28, 2025, Duolingo CEO Luis von Ahn sent a company-wide email, later shared on LinkedIn, announcing that Duolingo would become an “AI-first” company. The plan included phasing out contractors wherever AI could do their work, and factoring AI usage into employee performance evaluations. Von Ahn framed the shift in stark terms: the company would “rather move with urgency and take occasional small hits on quality than move slowly and miss the moment.”
The backlash was immediate and unusually sharp for a language app. Users who had built multi-year streaks posted about deleting the app in protest. Comments accumulated across social media with a consistent theme: “AI first means people last.” Subscribers complained that lesson content had visibly declined in quality, with one describing it on Reddit as “AI slop.” Duolingo’s stock, which had peaked above $540 a share in May 2025, had fallen roughly 80% from that high by April 2026.
Von Ahn later told the Financial Times he “did not expect the amount of blowback.” Within about a week of the original announcement, he walked back the framing, clarifying that he did not “see AI replacing what our employees do,” but rather viewed it as “a tool to accelerate what we do, at the same or better level of quality.” By April 2026, the company had quietly dropped AI usage from employee performance reviews altogether, reportedly after employees themselves began asking whether they were expected to use AI simply for its own sake.
Duolingo for Schools: A Separate but Related Casualty
The second development is more directly relevant to teachers. In April 2026, Duolingo announced that Duolingo for Schools, its free classroom management layer launched in 2015, would sunset on July 31, 2027. No new classrooms or accounts can be created going forward, though existing users can continue adding students to current classrooms until the shutdown date.
It is worth being precise about what is actually ending. Duolingo for Schools was a fairly thin wrapper: it let teachers assign existing Duolingo units to students and track whether they hit a completion or XP target. There was no custom content creation, no detailed error analysis, and no way to modify the underlying consumer library. Duolingo’s official explanation was that the tool did not see “satisfactory levels of use” relative to the investment required to maintain it, and the company’s core business model, built on individual subscriptions and in-app purchases, was never a good structural fit for a free institutional dashboard with roster management and compliance needs.
The regular Duolingo app is not going anywhere and will remain free for students and teachers to use exactly as before. What disappears is the assignment and progress-tracking layer on top of it.
Reading the Two Events Together
The AI-first reversal and the Duolingo for Schools shutdown are not the same event, but they rhyme. Both trace back to the same underlying tension: Duolingo is fundamentally a consumer app company, and both its AI content strategy and its classroom tool were attempts to stretch that consumer product to serve institutional and educational needs it was not originally built for. When the AI-generated content strategy produced a visible drop in quality, the company’s own paying users revolted loudly enough to force a public reversal. When the free teacher dashboard failed to justify its maintenance cost against a business model built on individual subscriptions, it was quietly discontinued.
For language teachers, the more useful lesson is not really about Duolingo specifically. It is that AI-first is not automatically better, and the market has started to notice. A large-scale meta-analysis published around the same time found that AI language tools do produce a meaningful positive effect on learning, but that effect was strongest in blended settings with a human teacher involved, and unreliable in online-only contexts (see our companion post on AI speaking practice tools for more detail). Duolingo’s own experience with AI-generated course content is a real-world version of the same finding: removing the human element from content quality control produced content users could immediately tell was worse, not just different.
What to Do If You Use Duolingo for Schools
The shutdown date is over a year away, which is enough time to make a considered transition rather than a panicked one. A few practical steps:
Decide whether you actually need a management layer. If Duolingo was mainly supplementary practice alongside a primary curriculum, and you were not relying heavily on the progress tracking, the honest answer may be to do nothing. Students can keep using the free Duolingo app exactly as before; you simply lose the assignment dashboard.
If you do need classroom management, know the realistic alternatives. Ellii (formerly ESL Library) offers a ready-made content library aimed specifically at adult and young adult learners with more robust reporting than D4S ever provided. Mango Languages is the closest institutional replacement, with a pre-built, standards-aligned curriculum, an admin portal, and LMS integrations, priced from roughly $6.67 a month per learner depending on the plan. Busuu for Education works well for secondary and university contexts that already run an LMS and want a structured course library with native-speaker community feedback built in.
Export what you want to keep now. Any progress reports or records worth referencing later should be pulled before the shutdown window closes, even though the deadline is well over a year out.
Use the transition period to look harder at the alternatives. Since D4S still works today, this is a reasonable window to trial a replacement side by side with the current setup before switching over completely, rather than making the decision under deadline pressure next summer.
AI-assisted, human-edited.