A class-action lawsuit against Duolingo shows how some educational apps prioritize profit over student learning. The legal action claims the platform redesigned its app to slow down free users and introduced low-quality, AI-generated content to cut costs. For parents and educators, the case exposes a tension between commercial business models and effective learning.
What Happened
As we previously reported, Duolingo is facing a class-action lawsuit led by the law firm Levi & Korsinsky, LLP. The lawsuit alleges that in April 2025, the company replaced its "Hearts" feature with an "Energy" system to penalize successful learners. Under the old system, free users only lost a heart when they answered incorrectly. The new system allegedly docked Energy even when users answered questions correctly. According to the lawsuit, this change was an artificial barrier to push free users toward paid subscriptions. The lawsuit also alleges that Duolingo's shift toward an "AI-first" strategy degraded lesson quality. Subscribers of the premium "Max" tier reportedly encountered translation and pronunciation errors during lessons, including conversations with the app's AI-generated avatar, Lily.
The Bigger Picture
This conflict is common in the digital learning market. A systematic review published in Postdigital Science and Education explains that many EdTech companies rely on "freemium" and subscription models. These commercial frameworks often prioritize growth and data collection over educational methods, which can create digital barriers for students from lower-income backgrounds. Consistent practice is hard to maintain. A study from EdWorkingPapers notes that keeping students engaged with digital learning tools is difficult, and added friction can stall progress. When platforms artificially limit usage, they disrupt the habits students need to succeed.
Relying on AI to generate educational content also carries risks. A study in PLOS One found that generative AI language tools have only "moderate reliability" in evaluating pronunciation and frequently over-score student performance compared to human teachers. Researchers writing in the International Journal of Business and Management also warn that unchecked AI adoption in lesson design often leads to "content bias" and a lack of depth, as human designers may uncritically accept automated drafts. As instructional expert Todd K. Edwards notes in a guide on Pedagrade, AI-generated courses frequently focus on simple rote testing instead of providing realistic practice and feedback.
What This Means for Families
For families and teachers, this shift means popular learning apps may not prioritize the student. When gamified apps use design tricks to pressure users into upgrading, learning outcomes suffer. As we noted in our coverage of how gamification can hurt learning, monetization mechanics distract from comprehension. The transition to AI-generated content also means lesson accuracy is not guaranteed. Parents who rely on mobile apps to help children learn a new language should know that digital tools cannot fully replace human instruction. Without adult oversight, students may learn incorrect pronunciations or memorize repetitive, low-quality phrases.
What You Can Do
First, parents should monitor app mechanics closely. Pay attention to how the app responds to a child's progress. If a platform begins penalizing correct answers or cutting sessions short to demand payment, consider switching to tools with transparent, non-punitive access.
Second, verify automated feedback with human help. Since generative AI can struggle with nuance and systematically over-score pronunciation, supplement app-based learning with human conversation, library books, or guided lessons from a qualified teacher.
Finally, supplement digital lessons with offline practice. Balance screen time with offline writing and speaking exercises. Actively discuss what your child is learning to make sure they are acquiring real-world communication skills, rather than just clicking buttons to satisfy an app's algorithmic constraints.