Libby is preparing to give readers more say over what they see as AI-generated books become harder to ignore. The change, described as “AI content controls,” is being positioned as a practical response to a problem that’s already showing up across digital publishing: the volume of new titles is rising, but the meaning of “new” is shifting. When books can be produced faster, at lower cost, and in greater variety—sometimes with little transparency about how they were made—library platforms face a new kind of curation challenge. Not just whether a title exists, but whether it should be surfaced by default, and how much control readers should have when they want to avoid certain kinds of content.
The move comes from OverDrive, the company behind Libby, which has long served as the front door for ebook and audiobook lending through tens of thousands of public libraries. OverDrive’s leadership is now treating AI as a major frontier rather than a distant trend. Marc DeBevoise, who recently took over as CEO, has framed AI as a disruption that will reshape the industry’s supply chain and reader experience. In that context, AI content controls aren’t being presented as a niche feature for early adopters; they’re being treated as infrastructure—something that needs to exist before the market becomes saturated with AI-generated material.
What makes this announcement notable isn’t only that Libby is planning controls. It’s the direction of travel: the platform appears to be moving toward a model where readers can actively manage AI-related content preferences inside the app, rather than relying entirely on publishers, distributors, or libraries to handle the issue upstream. That shift matters because library systems are not uniform. Different libraries have different budgets, different cataloging practices, and different tolerance for risk. If the burden of AI filtering falls solely on institutions, the experience will vary widely from one patron to another. If the burden shifts toward reader-level controls, the experience becomes more consistent—at least within the same app.
At the center of the story is a simple question that’s becoming increasingly difficult to answer in modern publishing: what does it mean for a book to be “AI-generated,” and how should that label affect discovery? For years, libraries have relied on metadata—author names, publication dates, genres, subject tags—to help patrons find what they want. But AI complicates metadata in two ways. First, AI can be used in the creation process without necessarily changing the outward packaging of a book. Second, even when a title is clearly AI-generated, the reader’s preference may not be binary. Some people may want to avoid AI-generated books entirely. Others may be fine with them if they’re well written, properly edited, or transparently labeled. Still others may want to explore AI-generated fiction as a cultural artifact, the way some readers treat fan fiction or experimental writing.
That’s why “controls” is a more meaningful word than “filter.” A filter implies a single outcome: show or hide. Controls imply a spectrum: allow, reduce visibility, require confirmation, or separate AI-labeled items into a distinct category. The Verge report indicates Libby is preparing to introduce options that let readers select how they want AI-generated material handled. Even without seeing the final interface, the concept suggests a user experience designed around choice rather than blanket exclusion.
This is also where the library platform’s role becomes more complicated than it might seem. Libby doesn’t just host content; it mediates access. It decides what appears in search results, what shows up in recommendations, and what’s easy to borrow quickly. Those decisions shape reading habits. If AI-generated books flood the catalog, the default discovery mechanisms could unintentionally amplify low-quality or misleading titles—especially if they’re optimized for engagement or produced at scale. Reader controls are one way to counterbalance that amplification.
But there’s another layer: trust. Libraries are built on a promise that information access is guided by public values—fairness, education, and community standards. When AI-generated content becomes common, patrons may worry about authenticity, originality, and the reliability of what they’re borrowing. Even if a book is entertaining, readers may still want to know whether it was created by a human author, trained model, or some combination of both. Controls inside Libby can function as a trust mechanism: they let patrons align their reading experience with their own comfort level.
Still, the effectiveness of any AI content control depends on something that’s easy to overlook: labeling and classification. A platform can only filter what it can identify. That means Libby’s controls likely rely on metadata provided by publishers or distributors, or on signals derived from content submissions. If AI-generated books are inconsistently labeled—or labeled in ways that don’t map cleanly to reader preferences—then the controls may work unevenly. In other words, the feature is only as strong as the ecosystem feeding it.
This is where OverDrive’s position as a distributor becomes relevant. OverDrive sits between publishers and libraries, and it has historically managed the operational complexity of digital lending: licensing, catalog ingestion, DRM, and delivery. As AI enters the picture, OverDrive’s job expands from logistics to governance. It has to decide what categories it can support, what definitions it will use, and how it will communicate those definitions to readers. If the company wants controls that feel intuitive, it must translate a messy reality into clear options.
One unique take on this moment is to view it less as a “feature rollout” and more as an early blueprint for how mainstream library platforms might handle AI at scale. For years, the debate around AI in publishing has often focused on authorship rights, copyright, and the ethics of training data. Those issues remain important, but they don’t fully address the day-to-day experience of readers. Readers don’t just ask “Is this legal?” They ask “Is this for me?” and “Can I trust what I’m seeing?” AI content controls are a direct attempt to answer those questions inside the product.
There’s also a broader implication for how libraries might evolve their relationship with discovery algorithms. Many digital platforms rely on ranking systems that optimize for relevance, popularity, or predicted interest. If AI-generated books become abundant, ranking systems could start to behave differently—not necessarily because they’re biased, but because the input data changes. More titles means more competition for attention. If AI-generated books are produced in large quantities, they may dominate certain niches simply due to volume. Reader controls can act as a counterweight, allowing individuals to opt out of that dominance.
At the same time, it’s worth acknowledging that “avoid AI” is not the only possible goal. Some readers may want to avoid AI-generated content that lacks human editing or that appears to be produced primarily for SEO-like discoverability. Others may be concerned about books that mimic existing styles too closely or that present themselves as factual when they’re not. The challenge is that these concerns aren’t always captured by a single label like “AI-generated.” A title could be AI-assisted but still heavily edited by humans, or it could be AI-generated but transparently marketed as such. Controls that only distinguish “AI vs non-AI” may not satisfy everyone.
That’s why the most interesting part of the Libby plan is the idea of giving readers multiple ways to handle AI content. If the final implementation includes options beyond a simple on/off switch—such as reducing visibility, separating AI-labeled items, or requiring explicit confirmation before borrowing—then it could better match the variety of reader preferences. It would also create a feedback loop: if readers consistently choose certain settings, OverDrive and libraries can learn what kinds of controls actually matter.
Another question is how these controls will interact with library-specific catalogs. Libby is used through public libraries, and each library curates its collection. If a library chooses to include AI-generated titles, Libby’s controls could still allow patrons to manage their personal experience. That means the system could preserve institutional autonomy while still empowering individual choice. In practice, that could reduce friction between patrons and libraries. Instead of arguing about whether a library should remove certain titles, patrons can adjust their own settings.
However, there’s a potential downside: personalization can fragment shared experiences. Libraries are often valued because they offer a common cultural space—everyone can browse the same shelves, even if they choose different books. If AI controls become deeply personalized, two patrons using the same library could see very different catalogs. That might be acceptable, even desirable, but it changes the nature of discovery. It turns the library from a shared environment into a customizable one.
The industry context makes this shift feel urgent. Digital publishing is already dealing with an explosion of content formats and distribution channels. AI accelerates that explosion. It also lowers barriers to entry, which can be good for creativity but risky for quality control. When quality varies widely, platforms become gatekeepers by default. The question is whether gatekeeping will be transparent and user-driven, or opaque and algorithmic.
Libby’s approach suggests the company wants to make gatekeeping visible. By offering controls, it acknowledges that readers should understand what they’re opting into. That’s a subtle but important difference. A platform that silently filters AI content might be accused of censorship or bias. A platform that offers controls can frame the decision as user agency. Even if the underlying classification is imperfect, the user-facing design can build legitimacy.
There’s also a policy dimension. Governments and regulators are increasingly interested in AI transparency, labeling, and consumer protection. While the details vary by region, the direction is clear: consumers should be informed when AI is involved in content creation, and they should have meaningful choices. Library platforms are not always at the center of these debates, but they are consumer-facing distribution channels. If Libby implements AI content controls in a way that aligns with emerging expectations—clear labels, understandable options, and accessible settings—it could become a reference point for other platforms.
For OverDrive, the business stakes are real. If patrons lose trust in the quality or authenticity of borrowed content, they may churn to alternatives. If libraries worry about reputational risk, they may hesitate to license certain categories. If publishers