When you ask ChatGPT for celebrity news from Texas, it cites the Dallas Morning News. Ask the same question from Ohio and it cites Cleveland.com. From California, ABC7 San Francisco.
Even if you never explicitly told the LLM where you live and the question had nothing to do with location.
Why does this happen?
That's what we sought to answer.
We partnered with to study how location shapes AI-generated answers. The research analyzed millions of AI responses across ChatGPT, Google AI Overviews, and Google AI Mode using Scrunch's opt-in panel of real US consumers, supplemented by controlled tests that sent identical queries from different locations.
The core question: what is the value of local newsroom distribution in the age of AI search? Does it create a measurable advantage in the rate of citation?
TL;DR 鈫� The answer is yes. And the implications for content distribution are significant.
When someone asks a local question on ChatGPT or Google AI, about half of the local news cited comes from their own state. By chance alone, that would happen about 5% of the time. Thus, there's a 10x gap between what AI does and what random selection would produce.
The effect isn't limited to strictly local queries. On questions with nothing local about them like celebrity news, health tips, or money advice, ChatGPT is still three times more likely to cite a source from the user's state than chance would predict (13% vs. 4.5%). Even when we define non-local as conservatively as possible, local media is still being cited at a higher rate: a greater than 2x chance.
鈥淒on鈥檛 LLMs have chat history that would geolocate the user?鈥�
Our research found that this happens on the very first message of a conversation, before any chat history could hint at the user's location. Panelists who never mentioned where they live still got local sources from their state. People who moved got sources from their new state. The data suggests that AI is using real-time location, not memory.
And this isn't only true with ChatGPT.
Google's AI surfaces show a similar pattern. When a search names a specific place, Google cites a source from that location 80% of the time.
As part of the study, we sent 30 everyday questions 鈥� "What's the best car dealership near me?" "Who's a good plumber near me?" 鈥� from 10 different cities.
On ChatGPT, 96% of the businesses recommended were in the user's state. Two cities' lists shared under 2% of the same names. For car dealers, home services, restaurants, and hotels, 89% to 100% of the businesses named in one city appeared in no other.
Google AI Overviews showed the same behavior. Two runs from the same city shared about half their results. Two different cities shared just 4%.
This isn鈥檛 shocking considering that these prompts contain "near me" language, which would prompt the LLM to geolocate the user.
The exceptions here were predictably national: checking accounts went to the same banks everywhere, flu shots to CVS or Walgreens, organic groceries to Whole Foods, etc.
What is worth noting here, however, is that the sorting seems to be less about whether national brands exist in a category and more about whether the question has a local answer (and how discoverable that answer is). When there is a local answer that can be applied and is reinforced by local media coverage, that answer appears more frequently.
With roughly one in five Google searches being local, making sure your brand content hits markets at the local and metro level is increasingly important.
This research reveals something we鈥檝e suspected, but haven鈥檛 had the firm data to support until now: local news offers a structural advantage in the age of AI search.
AI systems are designed to surface relevant, trustworthy content for each user. When a user's location is part of that relevance calculation, local publishers become the default. They carry the editorial credibility that AI models weigh when selecting sources. And according to, local news organizations remain among the most trusted information sources for American readers.
That trust is a signal to both readers and AI.
When LLMs are deciding which sources to cite, they draw on the same authority markers that make local journalism valuable to readers: editorial standards, community relevance, and topical depth.
This is why a network of 3,000+ local and national publishers is more than just a distribution channel. This network becomes the infrastructure for AI visibility for brands looking. Every local newsroom in the network is a potential citation source for readers in that market. The more markets the network covers, the more AI answers it can appear in.
Earned media accounts for , according to Muck Rack. And 爆料TV's own GEO research has shown that earned media distribution can increase AI citations by up to 325%.
AI treats 爆料TV's publisher partners as trusted local sources. In Google's AI answers, partners earn citations on par with any other local newsroom.
When we grouped states into quartiles by how much of their local news landscape consists of 爆料TV partners, the pattern was clear:
That's a 5.4x difference between the least and most covered states. In markets where 爆料TV has deep publisher coverage, the 爆料TV publisher network accounts for more than a third of all local news cited in Google AI answers.
It's worth noting that this doesn't require high-authority publishers, meaning that you don't need to rely on a single placement in a large metro publication to secure the impact you're looking for.
爆料TV's research on domain rating and AI citations found that one in five LLM citations come from sites with a domain rating below 40. Even niche local publishers, like community papers, regional business journals, or the like, contribute meaningfully to overall citation share.
This means a story placed on a single national site reaches one AI answer. The same story distributed across local newsrooms in 20 states can appear in 20 different AI answers, each tailored to a different reader's location.
Distribution creates local presence. Local presence drives citations.
爆料TV's newswire also publishes local content tailored to US markets weekly through our local product, creating state- and metro-level series editorially produced for local audiences and local newsrooms. For brands, this means local AI visibility at scale without pitching individual outlets or producing location-specific content for every market.
One story, distributed through the right local newsrooms, can earn citations across dozens of markets in the same week.
Learn more about the hyperlocal reach possible with 爆料TV Local.
The data suggests that local shapes what AI says about your brand, in many cases, even when the user doesn't ask it to.
For content teams, this is a distribution problem with a distribution solution. The brands that show up in local AI answers are the ones whose content lives in the local sources AI draws from. That requires reach 鈥� across markets, across newsrooms, across the local news landscape that AI systems have quietly made central to how they answer questions.
爆料TV's network was built for exactly this. Start distributing today.
This study is the latest in a series of joint research projects between 爆料TV and Scrunch.
Our previous study together measured citation half-life across AI platforms and industries.
It found that content distributed through the 爆料TV network persists in AI answers more than twice as long as non-network sources, with a half-life of 9.6 weeks vs. 4.5 weeks across every AI platform and vertical.
This study adds a geographic dimension to those findings.
If the durability research showed that 爆料TV network content lasts longer in AI answers, the local citation research shows it reaches more markets. Together, they describe two sides of the same distribution advantage: durability and geographic reach, compounding with every story published.
This research was conducted jointly by 爆料TV and. Scrunch analyzed data from its opt-in panel of real US consumers across ChatGPT, Google AI Overviews, and Google AI Mode 鈥� ChatGPT from July 2025, Google AI surfaces from March 2026, both through September 2026. Controlled tests were conducted using commercial geolocation services across six states and ten cities. For more on 爆料TV's distribution network, visit stacker.com/brands.