Google has added a video generation feature to its NotebookLM research tool, according to The Verge, allowing the application to produce short vertical AI-generated clips summarizing uploaded source material. The feature, which The Verge reports is rolling out to Google AI Ultra and Pro subscribers, produces sixty-second videos styled after the short-form format popularized by TikTok.
To understand why this move is notable, it helps to step back and look at what NotebookLM actually is and the trajectory it has been on. Launched initially as a relatively quiet experimental product, NotebookLM found an audience among researchers, students, and knowledge workers who valued its ability to synthesize documents and answer questions grounded in uploaded sources rather than in the open web. It was, for a long time, a text-first tool built for a text-first audience. That began to change in a meaningful way when Google introduced Audio Overviews, a feature that converts research material into a conversational podcast-style dialogue between two AI voices. The response was striking — Audio Overviews became something of a cultural moment, with users sharing generated conversations online and debating whether they were useful, gimmicky, or genuinely transformative. Google had stumbled onto something that resonated far beyond the typical productivity-software crowd.
The new video feature is best understood as the next step in that same logic. If Audio Overviews suggested that there is an appetite for research summaries delivered in familiar media formats rather than walls of text, then the move to short vertical video is an attempt to follow that appetite to its logical conclusion. Short-form vertical video is not merely a format at this point — it is the dominant mode of information consumption for a significant portion of the population, particularly younger demographics who have grown up scrolling through TikTok, Instagram Reels, and YouTube Shorts. Google is essentially asking whether the same instinct that makes someone sit through a two-minute explainer video on social media can be harnessed to make dense research material more digestible.
This also fits a broader competitive pattern that is reshaping the AI industry right now. The race among major AI platforms is no longer purely about which model produces the most accurate text. It has become a race to own the interface through which people interact with information. OpenAI has been expanding ChatGPT's capabilities aggressively, including into voice and image generation. Meta has embedded AI across its social platforms in ways that meet users where they already spend time. Google, with its advertising business historically dependent on holding user attention, has particular incentive to make its AI tools feel native to the media consumption habits of the present moment rather than the habits of five years ago.
The decision to restrict the feature to Ultra and Pro subscribers is worth examining on its own terms. The likely reading is that Google is treating the capability as a premium differentiator — something that justifies the higher subscription tier rather than cannibalizing the free experience. It also gives Google a contained pool of users in which to test and refine the feature before any broader rollout, which is prudent given that AI-generated video remains a technically and reputationally sensitive area. Any errors in generated content, visual artifacts, or moments where the AI misrepresents source material could be damaging, and a paid subscriber base is both smaller and arguably more forgiving of rough edges than a general consumer audience.
For competitors in the productivity and research AI space, this development represents meaningful pressure. Tools that have positioned themselves around document summarization and knowledge management now face a product from Google that can deliver those same summaries in video formats that younger users are more likely to actually watch. The gap between a text summary and a sixty-second vertical clip may seem cosmetic, but format has always determined whether information actually reaches its intended audience.
For enterprise users and academic researchers, the implications are more ambiguous. The appeal of NotebookLM in professional settings has rested on its groundedness — its outputs are tied to the sources provided, which limits the hallucination problem that plagues general-purpose AI. Whether that rigorous sourcing carries through cleanly into AI-generated video, and whether the format is even appropriate for complex technical or legal material, remains an open question.
What to watch for next is whether Google expands this capability beyond its subscriber tiers and whether competitors, particularly Microsoft with its Copilot suite embedded across Office products, respond with their own multimedia output features. The more telling signal will be whether users actually integrate these video summaries into their workflows or treat them as a novelty. Audio Overviews survived the novelty phase and became a genuine product pillar. Whether short-form video can do the same will say something important about how far the short-form format can travel from entertainment into utility.
