Yahoo is trying something very unfashionable in 2026: it wants you to click on links again. At a time when every AI search demo seems obsessed with keeping you inside a single answer box, Yahoo’s new AI product, Scout, feels almost retro — in a good way. It looks like an AI search engine, acts like an AI search engine, but quietly makes a different bet: that the future of search isn’t just answers, it’s a better relationship between AI and the open web.

If you haven’t tried it yet, Scout lives as a tab inside Yahoo Search, as a standalone site at scout.yahoo.com, and inside the revamped Yahoo Search app on iOS and Android in the US. You get the familiar big text box, suggested prompts, and a conversational answer when you ask a question — exactly the vibe you’d expect if you’ve used Perplexity or Google’s AI mode. Under the hood, Scout is powered primarily by Anthropic’s Claude model, grounded with Microsoft Bing’s search index, and then layered with Yahoo’s own data, content, and personalization. Think of it as Yahoo taking a foundation model off the shelf, then wiring it into 30 years of search and media infrastructure instead of trying to reinvent the whole stack.

What makes Scout interesting isn’t the model choice — Claude plus Bing is quickly becoming a very normal enterprise combo — it’s the way Yahoo has decided to present the answers. When you ask something like “What’s the latest on this winter storm?” Scout gives you a tight summary at the top, but it doesn’t stop there. The answer is flanked by clearly visible blue links, a “Latest news” block, and a full list of sources you can expand to see everywhere it pulled from, including Yahoo’s own verticals and other publishers around the web. On one test query about local weather, the response page surfaced nine links in total — far more than you typically see foregrounded in AI modes from Google or Perplexity, where citations often live behind icons or tucked-away buttons.

This is deliberate positioning. Yahoo’s executives talk about Scout as an “answer engine,” but also as a kind of AI-native reboot of the old portal idea — a guide that helps you navigate a messy internet rather than a chatty assistant that tries to replace it. Eric Feng, who leads Yahoo’s research group and the Scout project, frames the problem as moving from “how do I find things on the internet” to “how do I sift through clickbait and AI slop.” The pitch is that Scout uses AI to compress the junk and surface the good stuff, without trapping you in a walled garden.

In that sense, Scout is weirdly on brand for Yahoo. The company started life in the mid‑90s as “Jerry’s guide to the World Wide Web,” a curated directory of sites designed to help people find worthwhile pages in the early sprawl of the internet. Search eventually ate that model, and Yahoo spent the next decade bouncing between being a portal, a media company, and a half-hearted search player while Google printed money. Scout is Yahoo’s attempt to fuse those identities: media company, search front end, and AI layer, all stitched together.

Crucially, Yahoo actually has the content and data to make this work. It owns big destinations like Yahoo News, Sports, and Finance, runs its own newsroom, and partners with a wide network of publishers. That means Scout is not just hallucinating off random webpages — it can lean on a “trusted” pool of stories, stats, and structured data, then blend that with Bing’s broader index of the open web. Yahoo says Scout is informed by more than 500 million user profiles, a knowledge graph spanning over a billion entities, and roughly 18 trillion consumer events each year across its products. That data helps it personalize responses and suggestions: think sports recaps tuned to your teams, finance answers that understand the stocks you follow, or shopping queries that remember your past interests.

At the same time, Yahoo is very intentionally not building its own foundation model. CEO Jim Lanzone has been clear that training a top-tier model is both expensive and not where Yahoo thinks its leverage is. Instead, the strategy is to “best serve users” with grounding data, content, and personalization layered on top of other people’s models — in this case, Anthropic’s Claude, plus Microsoft’s infrastructure for indexing and grounding. That puts Yahoo in the camp of companies betting on orchestration and product design, not raw model horsepower, as the way to stand out in AI.

There’s also an unusually pragmatic story about business incentives here. Because Yahoo doesn’t have a massive, fragile search ads empire like Google’s, it can move faster in making AI the default search experience without worrying about cannibalizing tens of billions of dollars overnight. Lanzone has hinted that Scout will eventually become the face of Yahoo Search, even if the classic list of “10 blue links” doesn’t disappear immediately. The money, at least in version one, comes from two places: affiliate links in shopping-style results and ad units placed low on the page to keep the experience relatively clean. Yahoo’s line is that this should be enough to keep Scout free for mainstream users, with a possible paid tier later if there’s appetite for more advanced features.

Where Scout really leans into Yahoo’s portfolio is everywhere outside the search box. The same AI layer is being wired into Yahoo Mail, News, Finance, Sports, and other services via what the company calls the Scout Intelligence Platform. In Mail, Scout can summarize long threads and pull out action items like dates or travel details. In News, it highlights key points from huge articles and can generate digestible overviews with links back to full stories. In Finance and Sports, it can answer questions about stock moves or game stats, sitting on top of live data feeds and editorial coverage. It’s the same idea everywhere: AI as a navigational layer that sits between you and Yahoo’s content, rather than a chatbot living in its own separate corner.

The other audience Yahoo cares about here is publishers. Every AI search player right now insists it cares deeply about the open web, but Scout is built to show that on the surface, not just in blog posts. Answers include inline citations and clearly labeled source links as a primary UI element, not a nice-to-have hidden behind an icon. Yahoo is also participating in Microsoft’s Publisher Content Marketplace pilot, which is meant to help funnel traffic and revenue back to content creators whose work gets used to train and ground AI answers. Longer term, Yahoo says it wants to give publishers access to impression and click data from Scout — potentially via a Webmaster Tools‑style dashboard — so they can see how their content is being surfaced in AI results.

Strategically, this is Yahoo trying to turn one of AI’s biggest PR problems — the perception that it freeloads on other people’s work — into a differentiator. If Google and newer players like Perplexity are seen as extracting value from the open web, Yahoo wants to be the company loudly “rebuilding the social contract” between search platforms and publishers. That narrative alone won’t win users, but it might matter for the ecosystem conversations brewing around licensing, fair use, and the future of news traffic in an AI-first world.

On the user side, Scout tries to hit a balance that a lot of AI search interfaces still miss. It doesn’t lean into the AI buddy persona or pretend to be your life coach; responses are fairly straightforward, almost old-school search in tone, just delivered in paragraphs and bullet points instead of a ranked list. The interface is bright and a bit playful — Axios notes colorful emojis in the sidebar, inline tables and images, and very obvious citations — but it’s still clearly about getting you to an answer fast. In early hands-on testing, Scout consistently felt more like “search with a conversational UI” than “ChatGPT with a URL bar,” which will probably suit people who just want to know when the Winter Olympics start without negotiating a multi-turn conversation.

Of course, Yahoo is not operating in a vacuum. It is stepping into a brutally competitive space where OpenAI, Google, Microsoft, and a stack of specialized AI search startups are moving aggressively. Google is slowly folding AI Overviews into core search, while trying not to blow up an ad machine that still funds most of Alphabet. Perplexity is chasing the “AI-native search” narrative hard, even as it takes heat from publishers over content reuse and attribution. Scout doesn’t obviously outgun any of these on pure capability; Claude is strong, but so are the latest models from OpenAI and Google.​

What Yahoo is really betting on is taste — that there’s a large group of people who don’t actually want search to become a full-blown AI assistant, but do want something smarter than 10 blue links. That cohort wants fast, trustworthy summaries, clear sources, and a path to click through when they’re ready to go deeper. If Scout can consistently feel like “classic search, upgraded” rather than “AI experiment,” Yahoo might have a real shot at clawing back relevance from the margins of the search market.

There are still a lot of open questions. Personalization at the scale Yahoo is promising will inevitably raise privacy and data‑use concerns, especially when it’s powered by 500 million profiles and trillions of behavioral events. The company will need to prove that its safety and alignment story matches the expectations it’s setting by choosing Claude for its judgment and guardrails. And then there’s the monetization tightrope: users clearly prefer cleaner, less ad-heavy AI interfaces, but the economics of AI search are still unproven compared to plain old ads on plain old links.​

Still, there’s something oddly refreshing about Yahoo’s move here. Instead of trying to wow you with pure AI spectacle, Scout is quietly opinionated about how search should feel: familiar, legible, and undeniably web-first. It wants you to read the answer, see where it came from, and then go visit the sites that made it possible. In a year when everyone seems to be racing to build an AI that replaces the browser, Yahoo is making a pitch for an AI that remembers why the browser mattered in the first place.