Google’s AI Mode in Search has taken another step toward transparency by adding richer inline source links directly inside AI-generated answers. Instead of a single cluster of citations at the bottom, users now see underlined terms and small external‑link icons woven throughout the text, each pointing to a specific source. This design change is meant to make it easier to see where particular facts or claims come from in real time.
The update lands at a pivotal moment. AI Overviews and AI Mode have moved from Google’s experimental Labs into default Search, reaching hundreds of millions of users at once. At the same time, regulators, publishers, and SEO professionals are asking tough questions about who benefits when AI sits between people and the open web. Richer source links are Google’s latest answer to that question, but whether they truly solve the “AI vs. publishers” tension is still up for debate.
How Google’s New Inline Source Links Work in AI Mode
Google’s latest rollout makes source links an integral part of the AI Mode reading experience. Inside an AI-generated paragraph, certain phrases are underlined; clicking them opens a new Google search for that term. Right next to many of those terms, a tiny external‑link icon appears, which jumps directly to a publisher page that informed that portion of the answer. The result is a dense network of navigational hooks sprinkled through each AI response.
This dual‑link pattern, inline search terms and adjacent external icons, was first formalized in AI Overviews, the technology that underpins AI Mode. Now, it is more visible and more consistent across AI Mode experiences. Google frames this as a way to reveal “how AI Mode got this answer,” connecting users not only to websites, but also to follow‑up search journeys they might not have considered on their own.
Crucially, the inline links sit alongside other attribution features like source carousels and side panels. Instead of relying on a single citation block at the end, Google distributes links at the sentence and phrase level. That makes it easier to trace individual claims, but it also keeps users inside the AI interface by default, with Google’s own search results still serving as a central hub.
From Labs to Default Search: AI Mode Goes Mainstream
For most of 2025, AI Mode lived in Search Labs, accessible mainly to early adopters and Google One AI Premium subscribers. In December 2025, Google flipped the switch: AI Mode, powered by Gemini 2.x, rolled into mainstream Search as a default experience in many regions. This shift means the updated source‑link behavior is no longer a niche experiment, it’s what everyday Google users now encounter when they type complex or exploratory queries.
As AI Mode emerged from Labs, its visual patterns also stabilized. Users now routinely see a multi‑source carousel or side panel next to the AI answer, with favicons, site names, and short snippets from several publishers. The new inline underlined terms and link icons complement this side display by tying specific sentences in the AI text to particular sources in that panel or beyond.
Because AI Mode is now widely available, its approach to attribution can shape user expectations of what “normal search” looks like. Instead of scrolling a simple list of blue links, people are increasingly primed to scan an AI summary first, then decide whether to click through. The mainstream rollout therefore raises the stakes: if inline source links are confusing, too subtle, or insufficiently clicked, the impact will be felt across the broader web ecosystem.
Google’s Stated Goals: Transparency, Exploration, and Attribution
Officially, Google presents the new source‑link design as part of a broader move toward more transparent AI answers. The company says expanded inline attributions help people “more easily explore topics and discover relevant websites,” giving clear signals about which parts of the AI response came from which sources. That goal resonates with long‑standing publisher demands for visible credit, not just abstract references hidden behind an AI interface.
Google also positions AI Mode as an assistant to web exploration, not a replacement. In public statements responding to publisher backlash, the company argues that engagement with the open web remains “healthy,” citing internal metrics that show users still click links after interacting with AI summaries. The richer inline links and source carousels are presented as evidence that AI Mode is built to support, rather than cannibalize, publisher traffic.
From a user‑experience perspective, the new pattern makes sense. Many people simply want a quick, coherent overview backed by trustworthy sources. Inline links that highlight where the AI is drawing from can increase user confidence, especially when dealing with sensitive topics like health, finance, or public policy. The challenge is ensuring that this design not only builds trust in the AI, but also sustains the publishers whose content makes those answers possible.
Regulatory Pressure and the EU’s Antitrust Probe
The timing of Google’s richer source links is not accidental. In December 2025, the European Commission opened an antitrust investigation into whether Google misuses publisher content for AI Overviews and other AI features without fair compensation or meaningful opt‑out options. Regulators are specifically probing how AI answers affect traffic and revenue for publishers, and how prominently their content is linked when AI sits at the top of the results page.
That scrutiny appears to be a key driver behind Google’s push toward more explicit, observable attribution in AI Mode. Inline underlined links and direct external icons are easier to showcase to regulators as tangible proof that Google surfaces sources clearly. For the Commission, however, visibility is only part of the story. The deeper question is whether these links actually deliver comparable economic value to the old top‑10 blue‑link model.
As the investigation unfolds, Google’s product design choices are being closely watched as signals of intent. By emphasizing multi‑source panels, inline citations, and dedicated link icons, the company can argue that AI Mode encourages a healthy ecosystem where users can pivot quickly from AI summaries to original sources. Whether regulators accept that narrative still depends heavily on measured traffic outcomes, not just UI screenshots.
How Inline Links Change the Web Traffic Equation
Early data from the Generative Engine Optimization (GEO) project paints a mixed picture of how AI citations affect publisher traffic. When AI Overviews or AI Mode show citations, cited websites often see around three times more brand‑name search queries, people searching for the site or brand after seeing it referenced. However, these same sites receive about 70% less direct click‑through compared to when they appear as traditional top‑10 organic results.
This suggests that richer source attribution increases awareness but doesn’t fully compensate for the “answer on the page” effect. Users may recognize and remember a brand without feeling a strong need to click through, especially when the AI summary already satisfies their informational needs. Inline links, even when clear and plentiful, cannot change the basic fact that the AI is compressing multiple web pages into a single, instantly consumable overview.
The case of Wikipedia underscores this tension. Despite being the most frequently cited domain in Google’s AI answers, Wikipedia reportedly saw an 8% decline in human pageviews in 2025. That drop happened even as the site’s logo, name, and snippets became more visible inside AI answers. In other words, richer AI citations and inline links are not a guaranteed safeguard against traffic loss, particularly for sites that used to capture a large share of “quick fact” queries.
Who Gets Cited? AI Mode’s Sources vs. Classic SEO Rankings
Another crucial shift lies in which sites benefit from AI Mode’s source‑link behavior. Large-scale analyses of millions of AI Overviews and AI Mode citations in 2025 show that about 89% of cited URLs do not appear in the top 10 organic results for the same queries. More than half of AI Overview sources don’t show up in the top 10 at all. That means AI Mode’s citation engine draws heavily from pages beyond the usual SEO winners.
Academic and industry researchers have noted that Gemini‑based search, including AI Mode, often cites domains that rank much lower in popularity indices like Tranco compared to standard Google Search. This break from traditional SEO signals suggests that AI models are optimizing for different qualities, such as topical coverage, sentence‑level relevance, or training‑data alignment, rather than classic link authority alone.
For smaller or niche sites, this can be a rare upside. Being cited in AI Mode may expose them to audiences they never reached via standard organic rankings. However, visibility does not always equate to visits. If most users get what they need from the AI answer, the long tail of newly cited domains might gain recognition, but not necessarily a meaningful bump in traffic or revenue.
Concentration of Citations: Big Platforms Still Dominate
Even as AI Mode taps into a long tail of less‑popular sites, a relatively small group of platforms still captures a large share of the citations. A 2025 study of roughly 36 million AI Overviews and 46 million citations found that Wikipedia, YouTube, Reddit, Amazon, and various Google‑owned properties collectively account for about 38% of all sources cited. In AI Mode, these domains frequently appear as primary references with prominent link treatments and favicons.
This concentration reflects broader realities of the web: large platforms host vast amounts of structured, frequently updated content that AI systems find easy to ingest and summarize. Wikipedia offers concise, generalist overviews; YouTube provides how‑to videos; Reddit aggregates real‑world experiences and niche expertise; Amazon supplies product data and reviews. From an AI answer standpoint, these are rich, reliable reservoirs of content.
However, that same concentration worries smaller publishers and specialized sites that rely on search traffic to sustain their operations. If AI Mode’s inline links and carousels routinely point to a handful of dominant platforms, smaller voices may struggle to gain meaningful visibility, let alone clicks. The design of AI Mode’s source presentation therefore has real consequences for how diverse, and how competitive, the information ecosystem remains.
Multimodal Queries: Rich Responses with Links by Default
Source links are not limited to text‑only queries. With the integration of Gemini and Google Lens, AI Mode now handles multimodal prompts, questions that include images, screenshots, or photos, while still promising “a rich, comprehensive response with links.” When users upload a photo or snapshot, the AI not only interprets the image but also surfaces related web sources inline and in side panels.
This multimodal behavior reinforces that linking out is now a standard expectation for AI Mode, not an optional add‑on. Even when the AI can infer a lot from an image alone, Google still emphasizes external references: guides, product pages, reviews, or explainer articles that give users more context and depth than a single AI paragraph can provide.
For visual‑first experiences, shopping, travel planning, home repair, or fashion inspiration, the combination of images plus AI summaries plus inline links is particularly powerful. Users can quickly understand what they are seeing, then jump directly to relevant sites for purchasing, detailed tutorials, or further research. At the same time, this concentrates power inside Google’s interface: every step of the journey, from capturing the image to discovering sources, is mediated by AI Mode.
Google’s decision to add richer inline source links to AI Mode answers is more than a cosmetic tweak. It is a strategic bet that visible, granular attribution can satisfy users’ desire for trustworthy information, ease regulatory concerns, and somewhat reassure publishers that they are still in the loop. The new design makes it easier than ever to see and click the sources behind AI summaries, whether you are reading about a news event, comparing products, or decoding a screenshot.
Yet the underlying trade‑offs remain unresolved. Independent data shows that AI citations can increase brand awareness while still reducing direct traffic, and even the most heavily cited sites are not immune to declining pageviews. AI Mode’s citation patterns also redistribute attention, favoring both major platforms and an eclectic mix of lower‑ranked sites, without guaranteeing proportional economic benefits. As AI‑powered search becomes the default, the critical question shifts from “Are sources visible?” to “Can the open web thrive when so many answers are consumed without a click?” The future of Google AI Mode, and the broader web, will be defined by how well that tension is addressed.
