Two recent federal court decisions in the EdTech industry reach very different conclusions. On March 4, 2026, a Connecticut jury returned a $75.3 million verdict against Course Hero/Learneo, Inc. in Post University Inc. v. Learneo, Inc., finding the platform liable for stripping copyright management information (CMI) from academic materials, replacing it with its own branding, and monetizing the result. On July 14, 2026, the Central District of California granted summary judgment in favor of Quizlet in Barkley & Associates, Inc. v. Quizlet, Inc., dismissing all claims against the online study platform.
Both cases involve crowdsourcing of materials through user-uploaded educational content, platform monetization through subscriptions, and claims of intellectual property infringement. Yet they reached diametrically opposed results. The difference comes down to a recurring theme in platform liability law: whether the platform merely provided infrastructure for user-directed activity or instead became an active participant in the alleged infringement.
The Quizlet Order: User-Directed AI as a Shield
Barkley & Associates produces test-preparation materials for nursing students studying to become certified nurse practitioners. Quizlet is an online study platform with over 60 million users that provides AI-enhanced study tools, including flashcards, study guides, and practice tests. Quizlet’s platform is powered by large language models from OpenAI and Google. Quizlet does not build or train its own LLMs.
Barkley alleged that its copyrighted materials appeared on Quizlet’s platform and that Quizlet’s AI tools were “spitting out” its content. But both parties agreed that students, not Quizlet, uploaded Barkley’s materials to the platform.
Direct Copyright Infringement
The Court found that volitional conduct, the threshold requirement for direct infringement, rested with the users, not Quizlet. Quizlet’s AI-generated study guides and summaries operated only after users selected or supplied the underlying study material. The Court concluded that operating an automated system in response to user-directed inputs did not make Quizlet the legal cause of the alleged copying. Barkley’s screenshots of AI-generated outputs were insufficient because those outputs were generated in response to user-submitted prompts referencing user-created flashcard sets. The Court’s reasoning rested on its determination that the AI is the user’s tool and the user remains the volitional actor even when the AI produces the output.
Vicarious Infringement
The Court found no “substantial influence” over users’ infringing activity and no “direct financial interest” in that activity. Both paying and non-paying users could access the same content on Quizlet. Subscriptions unlocked greater frequency of AI tool usage, not access to any particular copyrighted material. This severed the causal connection between infringement and revenue that vicarious liability requires.
DMCA § 512(c) Safe Harbor
After concluding that Barkley had failed to establish direct liability, the Court separately held that, even assuming for the sake of argument that liability could otherwise exist, Quizlet independently qualified for DMCA safe harbor protection. The Court found Quizlet maintained an effective takedown process and removed 3,881 study sets in response to Barkley’s notices, although the opinion did not address how many takedown notices were submitted. It should also be noted, however, that Barkley raised arguments about unidentifiable users and “tiny URLs” for the first time at oral argument, which the Court deemed waived.
Trademark and Unfair Competition Claims
The Court dismissed Barkley’s Lanham Act claims on multiple grounds. The Court applied Dastar Corp. v. Twentieth Century Fox Film Corp. to bar Lanham Act claims that were functionally congruent with the copyright claims. Barkley also failed to establish common law trademark rights in the standalone word “Barkley.” Finally, the Court found the state Unfair Competition Law claim preempted by Section 301 of the Copyright Act.
The Post University Verdict: When the Platform Becomes the Actor
In Post University Inc. v. Learneo, Inc., the District of Connecticut presided over a nine-day jury trial that began February 17, 2026. The jury returned its verdict on March 4: $75,375,114.85 in Post University’s favor.
Course Hero’s business model was straightforward: Users uploaded academic materials—syllabi, assignments, quizzes, and exams—which Course Hero then processed and monetized. Unlike Quizlet, Course Hero did not simply host user content. The jury heard evidence that Course Hero affirmatively removed Post University’s CMI and replaced it with its own: logos, banners, watermarks, and copyright notices. The rebranded content was then locked behind a paywall and monetized through subscriptions and document “Unlocks.”
The jury found 1,458 violations of DMCA § 1202(a) (providing false copyright management information) at the statutory maximum of $25,000 per violation, totaling $36,450,000. It found 1,557 violations of DMCA § 1202(b) (removing or altering CMI) at the same statutory maximum, totaling $38,925,000.
The jury found that the copyright infringement damages (separate from the CMI issues) on five registered Works amounted to just $114.85. In doing so, the jury rejected Course Hero’s safe harbor and fair use defenses on all five registered Works.
The safe harbor finding is significant. Course Hero presumably maintained some form of takedown process, yet the jury answered “no” to safe harbor on every Work. The verdict form does not explain the jury’s reasoning, but the factual basis likely rested on Course Hero’s affirmative transformation of the content. Section 512(c)(1) protects “storage at the direction of a user,” and content that has been stripped of its original CMI, rebranded with the platform’s logos, and gated behind a commercial paywall is no longer being merely “stored” in any meaningful sense. When a platform’s conduct moves beyond passive hosting into active modification and commercialization, the statutory predicate for safe harbor may simply not apply, regardless of whether the platform also complies with takedown notices.
The Dividing Line: Platform Conduct as the Determinative Factor
Read together, the Quizlet order and the Post University verdict illustrate that liability turns on whether the platform merely provides the infrastructure through which users act or whether the platform itself becomes the actor that copies, alters, or commercializes copyrighted works.
Quizlet occupied the infrastructure end of the spectrum. Users uploaded content. Users invoked AI tools. Quizlet provided AI-enhanced study features that responded to user-selected inputs and complied with takedown notices. The Court found this insufficient to establish the volitional conduct required for direct infringement, the control and financial interest required for vicarious infringement, or any basis to defeat safe harbor protection.
Course Hero occupied the opposite end. It did not merely host content; it stripped away the original owner’s identifying information, substituted its own branding, and sold access to the resulting product. The platform was not simply a tool that users directed; the jury found that it acted upon the content for its own commercial purposes.
The Emerging Power of DMCA § 1202: CMI Stripping
One of the most significant takeaways from Post University is the independent force of DMCA § 1202 as a damages theory. Post University had only five registered copyrighted works, and the infringement damages on those works totaled a negligible $114.85. The overwhelming majority of the $75.3 million verdict came from statutory damages for CMI violations.
Copyright infringement claims require timely registration—at minimum, registration before suit is filed. Statutory damages and attorney’s fees for infringement require registration before infringement or within three months of publication.
Section 1202 protects against the removal or alteration of copyright management information—a category that includes titles, author names, copyright notices, and terms of use—and provides statutory damages of $2,500 to $25,000 per violation, compared to an overall range of $200 to $150,000 per Work for traditional copyright infringement. Where a platform systematically processes thousands of documents, as the jury found in Post University, the per-violation statutory damages can aggregate to staggering sums without a single copyright registration.
The type of legal theory available to a content owner depends on what the platform actually did with the content, not merely whether the content was present. A platform that hosts infringing material in its original form exposes itself to traditional copyright claims (subject to safe harbor and volitional-conduct defenses). A platform that strips or replaces CMI exposes itself to § 1202 claims carrying their own independent damages framework. The lesson is not simply that platform conduct determines liability; it determines which kind of liability is even on the table.
Implications and the Road Ahead
These decisions arrive as AI-powered tools increasingly interact with user-supplied content. Interested entities should be considering:
- Platforms that provide tools for users to interact with their own uploaded content—without modifying, rebranding, or selectively monetizing that content—are more likely to succeed on volitional conduct, vicarious liability, and safe harbor grounds.
- Platforms that affirmatively process, rebrand, or strip identifying information from uploaded content risk transforming themselves from tool providers into content actors, assuming the additional legal exposure that entails, including DMCA § 1202 statutory damages that require no copyright registration.
- AI-enhanced features do not, by themselves, create liability. The dispositive question for now is who initiated and directed the AI’s activity.
For content owners, the message is equally clear: The strength of a claim depends not simply on the presence of copyrighted material on a platform, but on what the platform itself did with that material.
