AI isn't just disrupting the publishing industry; it's fundamentally reshaping how audiences discover and consume news. For years, publishers relied on search traffic to power ad revenue, subscriptions, and engagement. However, the numbers now paint a stark picture: the old model is collapsing, and a new one is urgently needed.
The Great Traffic Migration
The numbers tell a stark story. Between January 2024 and May 2025, publishers lost over 600 million visits, a 25% drop in organic traffic that fell from 2.3 billion to 1.7 billion. Meanwhile, ChatGPT news queries more than tripled, up 212%.
It's not just about search behavior changing. Readers are fundamentally shifting where they look for answers. Google searches for news dipped 5%, but ChatGPT's user base, both app and web, grew more than 50%. Mainstream news consumers are now routinely turning to conversational AI instead of traditional search.
The Zero-Click Era
Adding to this pressure is the rapid rise of zero-click searches. Since the launch of Google's AI Overviews in 2024, the share of news-related searches ending without a single click jumped from 56% to nearly 69%.
This means that almost seven out of ten users now get their answers directly from AI-generated summaries, without ever reaching a publisher's website. For publishers who depend on ad impressions and subscription conversions, the economics no longer work. Traffic is down, revenue is down, and the traditional SEO playbook—keywords, backlinks, and click-optimized headlines—can't fix it.
From Browsing to Asking
Behind these numbers lies a more profound behavioral shift. Users no longer want to scroll through headlines or navigate complex sites. They want instant, contextual answers to direct questions.
Instead of typing "inflation news today" and browsing five different articles, users ask ChatGPT: "How is inflation affecting household spending this month?" The answers they receive combine context, synthesis, and explanation in a way that traditional search has never delivered.
Politics, climate, and the economy now dominate AI-driven queries, while traditional drivers, such as sports scores and weather updates, are declining. The news consumer is evolving, from a browser of headlines to a question asker.
The Hidden Asset: Decades of Archives
Yet amid this disruption, publishers hold an undervalued advantage: decades of archived content that AI platforms desperately need but can't easily access or understand.
Most publishers sit on massive libraries of video footage, audio recordings, transcripts, and documents, footage of historical events, expert interviews, on-the-ground reporting, and cultural moments that shape public understanding. However, this content remains underutilized and undermonetized, locked in storage systems, unsearchable beyond basic tags, and disconnected from modern AI workflows.
The irony is sharp: as AI systems struggle with accuracy and context, publishers have precisely what they need, authoritative, rights-controlled archives spanning years or decades. The problem isn't the content itself; it's that these archives exist in formats AI can't easily parse, query, or monetize.
Video: The Most Valuable (and Most Locked) Asset
Video archives represent the most complex and valuable content publishers own, and the most underutilized. Unlike text, which can be indexed and searched relatively easily, video remains opaque. A 30-year archive of news footage might contain hundreds of thousands of hours of content, but without frame-level analysis, it's essentially unsearchable.
Traditional video management systems rely on manual tagging; someone watches footage and adds keywords. This approach doesn't scale, misses nuance, and can't answer the kinds of questions AI users now ask: "Show me every clip where a politician discussed climate policy in 2019" or "Find footage of protests in major cities during economic downturns."
This is where Video Intelligence changes the equation. Rather than treating video as static files in a library, Video Intelligence transforms each piece of footage into structured, searchable intelligence. It analyzes every frame to identify scenes, objects, logos, people, and text. It generates transcripts, extracts entities, and creates what we call a "video dossier", a comprehensive breakdown of what's actually in the footage.
More importantly, it does this at archive scale, not just for individual clips. The system builds connections across thousands of hours of content, enabling semantic search that understands meaning and context. A query like "coverage of wildfires affecting California communities" doesn't just return clips with the word "wildfire", it surfaces relevant scenes based on visual analysis, transcript context, and thematic connections across the archive.
Rights-Aware Intelligence
One of Video Intelligence's critical differentiators is rights-aware metadata. Publishers don't just need their archives to be searchable; they need control over how that content is used, distributed, and monetized.
Video Intelligence embeds usage rights directly into the metadata layer. Administrators can set policies that automatically apply across queries, such as restricting clips with expired licensing agreements, filtering out logos of specific brands, excluding sensitive footage, or limiting distribution to certain geographic regions. This means publishers can safely expose their archives to AI platforms and partners without losing control or risking compliance issues.
For sports leagues, this might mean filtering out clips that violate broadcast exclusivity. For news organizations, it could mean ensuring violent or graphic content is flagged and restricted. For entertainment archives, it's about managing talent rights and union agreements.
From Storage to Revenue Streams
With Video Intelligence, what was once stored assets becomes a revenue stream. Instead of sitting on decades of footage that generates no return, publishers can:
- Deploy API endpoints that let partners query their archives directly, paying per query rather than per dataset
- Package thematic collections instantly, "every major political debate from the last decade" or "disaster response coverage across regions."
- License to AI training datasets, providing the high-quality, contextualized footage that AI models need to improve accuracy
- Power new editorial products, from automated highlight reels to AI-assisted research tools for journalists
The entire archive becomes monetizable and actionable inventory, not dead weight. And because the system operates at the query level, publishers maintain visibility into what's being accessed, how often, and by whom, giving them the data they need to optimize pricing, identify high-value content, and negotiate from a position of strength.
Why Traditional Models Won't Save Publishers
Publishers have experimented with AI licensing deals, but the math doesn't add up. Even large contracts, like $250M over five years for News Corp, are tiny compared to the billions being lost annually in traffic-driven revenue. And one-time "pay-per-crawl" models only compensate once, no matter how often AI systems reuse that content.
As Anthony Katsur, CEO of IAB Tech Lab, put it: "Pay per query scales. Pay per crawl does not."
The Path Forward: Activating Data Archives
There is an alternative. Instead of accepting one-off licensing fees, publishers earn revenue every time their content helps answer a question in an AI response.
This model ensures:
- Fair compensation: Publishers could get paid proportionally to how often their content is used.
- Control and transparency: Rights-aware systems ensure attribution, ownership, and traceability
- Scalability: as AI adoption grows, so does publisher revenue
It reframes the relationship between publishers and AI platforms from one of adversarial to one of symbiotic nature. AI systems gain the authoritative content they need, while publishers finally unlock sustainable revenue streams that grow in tandem with AI usage.
The Publisher's Choice
The question isn't whether AI will continue to reshape news consumption; that's already happening. The real question is whether publishers adapt quickly enough to survive and thrive in this new environment.
Those who cling to traffic-driven models risk accelerating decline. Those who embrace their archives as strategic assets, transforming dormant footage, audio, and documents into structured, searchable, monetizable intelligence, can turn existential threat into competitive advantage.
The content is already there. The technology to unlock it exists. What remains is the decision to act.