Google Launches AI E-Book Feature for Research with Copyright Protections

Google has officially launched a new AI feature designed to integrate commercial ebooks into its research tools, promising built-in protections for author and publisher copyright. The feature, named Expert Intelligence, launched within the tech company’s Gemini Notebook application, formerly known as NotebookLM, allowing users to draw directly on digital books to generate summaries, cross-reference sources, and answer complex research queries.

The initial rollout includes access to more than 100,000 titles across various academic, professional, and trade sectors. Major commercial and academic publishers participating in the launch include Bloomsbury, Macmillan Publishers, Penguin Random House, O’Reilly Media, De Gruyter Brill, and Johns Hopkins University Press. By bringing fully copyrighted texts directly into AI workspace tools, Google is attempting to create a workable middle ground between generative AI functionality and conventional intellectual property rights.

How the system handles copyright and access

The core mechanism of Expert Intelligence relies on verified individual ownership. To use a book as an active data source within Gemini Notebook, a reader must have already purchased that title through Google Play Books. The AI operates on top of the user’s existing digital library rather than scraping an open repository.

Crucially, digital rights management rules carry over into the collaborative workspace. If a user shares a workspace notebook containing analysis or citations from a specific ebook, collaborators who do not own that title will not be granted access to the underlying text or content derived from it. Google confirmed the details in an official social post, highlighting that the platform launches with titles spanning major trade and university presses.

This approach attempts to resolve a longstanding friction point in digital publishing. For years, the integration of long-form books into analytical tools has been stymied by fears of piracy or uncompensated data mining. If you ask me, this explicit paywall requirement offers a clear template for how tech platforms might licence commercial writing without bypassing traditional sales channels.

Publishing context and unresolved legal friction

The collaborative launch arrives during a period of intense legal scrutiny regarding tech platforms and copyright. While this individual purchasing model sets a structured precedent, it sits alongside broader industry tensions over how AI models were trained in the first place. Tensions surrounding digital scanning and training datasets have dominated industry discussions, from debates over how commercial texts are processed for machine learning to complex licensing disputes across major trade houses.

Indeed, Google remains locked in ongoing legal proceedings, including a federal lawsuit filed by authors and publishers alleging that earlier iterations of the Gemini model were trained on copyrighted works without explicit consent or compensation. The launch of Expert Intelligence does not resolve those past disputes, but it does mark a distinct shift toward a permissions-based framework for active reader tools.

For publishers, participating in structured AI models offers a potential new revenue stream at a time when digital sales strategies are undergoing rapid shifts. Recent industry reports show steady resilience across the market, with publisher revenues growing steadily as trade houses experiment with audio, subscription platforms, and direct-to-consumer digital channels.

What this means for readers and authors

For researchers, students, and non-fiction readers, the practical appeal of Expert Intelligence is straightforward. Gemini Notebook has earned a reputation as a helpful tool for organising complex notes and documents; adding verified, full-length books into that workflow allows for precise semantic searching across dense texts. Instead of relying on general web data or unverified summaries, users can interrogate trusted texts directly.

Whether this cooperative approach will fully satisfy author groups remains to be seen. Writers’ organisations have consistently argued that any use of copyrighted prose in automated systems requires direct opt-in permissions and fair royalty structures. While requiring a book purchase at the point of use ensures that a sale takes place, questions remain about how fair compensation is calculated when AI generates extensive synthesis from a writer’s work.

For now, the launch represents one of the largest official partnerships between major trade houses and a tech giant on AI reading tools. As the system expands beyond its initial 100,000 titles, the industry will be watching closely to see whether sales follow the technology.