Google announced this week that AI Mode—its AI-powered search feature—will include more inline links to sources and add AI-generated snippets explaining why those sources are relevant. The Verge reports that Google will place descriptions like "These articles suggest budget-friendly decor ideas, including secondhand shopping, architectural updates like molding and hardware swaps, and DIY projects to achieve a vintage look" directly above the carousel of source links.
The timing is conspicuous. The European Commission opened an investigation into Google days earlier, examining whether the company violates competition rules by using publisher content in AI features "without appropriate compensation." Google also pushed back in August on research claiming users are "less likely" to click links when AI Overviews appear, asserting that click volume has remained "relatively stable."
Now Google is adding more citations. Let's examine whether that addresses the actual problem.
Google's update puts source links directly inside AI-generated responses rather than relegating them to a separate carousel. It also adds AI-written explanations of why each source matters. In theory, this increases transparency and makes it easier for users to verify information or explore original content.
In practice, it reinforces the same dynamic publishers have been complaining about: users get the information they need from the AI summary and have no reason to click through. Adding a sentence explaining that certain articles "suggest budget-friendly decor ideas" doesn't create incentive to read those articles. It provides additional context that makes clicking through less necessary, not more.
Google claims click volume has remained stable despite AI Overviews. Publishers dispute this. Neither party has released comprehensive data. What we know for certain: AI Mode synthesizes content from multiple sources into a single answer, and synthesized answers reduce the need to visit original sources—that's the entire value proposition.
Linking to sources is not the same as compensating publishers for content use. Google's AI Mode ingests publisher content during training and inference, generates responses based on that content, and displays those responses in ways that satisfy user queries without requiring clicks. Adding more citations doesn't change the fundamental economic model: publishers create content, Google extracts value from it, and users consume that value without visiting publisher sites or generating ad revenue.
Google is also launching a pilot program with The Guardian, The Washington Post, and the Washington Examiner to "explore how AI can help drive more engaged audiences." One experiment involves displaying AI-written overviews of articles inside Google News. This is framed as partnership, but the structure is familiar: Google produces AI summaries of publisher content, reducing the need for users to read the original articles.
If AI-written overviews increase engagement, great. But engagement with what? With summaries Google controls, or with the original reporting those summaries are based on? The pilot doesn't clarify whether publishers will see increased traffic, increased revenue, or just increased visibility for content they no longer control the distribution of.
The European Commission's investigation hinges on whether Google provides "appropriate compensation" for content use. That phrase is doing a lot of work. Does appropriate compensation mean licensing fees? Revenue sharing? Attribution with no payment? The answer matters enormously, and nobody—including regulators—has defined it clearly yet.
If compensation means payment, Google faces a choice: negotiate with every publisher whose content appears in AI responses, or stop using that content. Neither option is operationally simple at the scale Google operates. If compensation means attribution, Google can argue they're already providing it through citations—even if those citations don't translate to traffic or revenue for publishers.
The regulatory gap here is that copyright and fair use weren't designed for AI systems that consume massive amounts of content during training and then generate derivative responses on demand. Courts will eventually decide whether this constitutes fair use or infringement. Until then, Google is optimizing for compliance with rules that don't exist yet.
Google's August statement that click volume remains "relatively stable" despite AI Overviews is technically possible and also potentially misleading. Stable click volume doesn't mean publishers aren't harmed if total search traffic increases while their share of clicks stays flat. It doesn't account for opportunity cost—the clicks publishers would have received if AI Overviews didn't exist.
It also doesn't distinguish between clicks on informational queries (where AI summaries are most useful and click displacement is highest) versus navigational or transactional queries (where users still need to visit specific sites). Aggregating across all query types obscures the actual impact on publishers who depend on informational search traffic.
Publishers have reported meaningful traffic declines. Google says click volume is stable. Both statements could be true if Google is measuring total ecosystem clicks while publishers are measuring individual site performance. The lack of transparent, granular data makes it impossible to adjudicate who's right.
Google's pilot with major publishers serves multiple purposes. It creates optics of collaboration rather than extraction. It generates data Google can use to demonstrate that AI features benefit publishers, not just Google. And it gives Google leverage in regulatory discussions by showing they're working with—not against—the content ecosystem.
But pilot programs with select large publishers don't solve the problem for the thousands of smaller publishers, niche sites, and independent creators whose content also gets synthesized into AI responses without compensation or meaningful attribution. The Guardian and The Washington Post have legal teams and negotiating power. Most publishers don't.
If Google's AI partnerships scale beyond a handful of prominent outlets, they could establish precedent for how content licensing works in an AI-mediated search environment. If they don't scale, they're window dressing—useful for public relations, irrelevant to the structural issue.
More inline citations don't fundamentally alter the relationship between AI search and publisher economics. They make attribution slightly more visible. They don't create incentive to click through, don't compensate publishers for content use, and don't address the core tension: AI systems that extract value from content while reducing traffic to the sources that created it.
Google is betting that improved attribution, pilot partnerships, and incremental interface changes will satisfy regulators and publishers. Publishers are betting that regulatory pressure and legal challenges will force substantive changes to compensation models. Users, meanwhile, are getting faster answers with less friction—which is exactly what AI search is designed to provide, and exactly what makes the economics unworkable for content creators.
The resolution—if there is one—won't come from interface updates. It'll come from courts, regulators, or new business models that haven't been invented yet.
If you're navigating platform dependency and need strategies that don't rely on traffic sources you can't control, Winsome's team can help you build leverage.