AI shopping's riskiest question: “Is there a store in my country?”

New data on 151 brands shows AI assistants sending shoppers to fake and dead stores, most often when they ask about a store in their own country. For any brand with regional domains, that's not an edge case.

Key facts

Podqi, a US brand protection firm, analyzed 6,795 answers from ChatGPT, Gemini and Perplexity about 151 brands in September 2026 using Gauge data; about 9 percent surfaced a risky site.
Risky sites were recommended 6.3 times more often than warned against; Perplexity showed them in 15, ChatGPT in 8 and Gemini in 4 percent of answers.
Questions about regional stores and international shipping were riskiest at 16 percent; of 718 risky domains, 351 were fake and 367 dead.

By Manfred „Luigi“ Lugmayr

Researched and written with AI, to Luigi's brief, positions and editorial rules. Automatically checked against the sources. How I work →

Picture a shopper asking ChatGPT whether a brand has a store in her country. According to a new analysis, that's the single riskiest question in AI shopping. And every global brand gets asked it daily.

What Podqi measured

Brand protection firm Podqi, working with data from AI search company Gauge, asked ChatGPT, Gemini and Perplexity five shopping questions about 151 brands in two rounds in September. That produced 6,795 answers and 4,437 individually reviewed domains.

Risk: about 9 percent of answers surfaced a risky site, meaning a suspected fake store or a dead domain. 718 such domains in total: 351 fake, 367 dead.
Recommend, not warn: 57 percent of those answers recommended the site, only 9 percent warned against it. A ratio of 6.3 to 1.
By assistant: Perplexity 15 percent, ChatGPT 8 percent, Gemini 4 percent of answers showed a risky site.
The country question: questions about regional stores and international shipping hit 16 percent, by far the highest. 36 percent of risky sites in those answers used a country-code ending or a region name.

For context: Podqi sells exactly the protection its study recommends. This isn't independent research.

The gap is your map, not the model

Here's what stays with me: the AI doesn't invent the fake stores. It fills a gap. If a brand never states clearly which country stores exist and which don't, the model takes whatever it finds. For any brand with a patchwork of regional domains, distributors in some markets and no store in others, that's an open flank.

My rule for AI visibility has been entity clarity before content volume. This study shows the flip side. Fuzzy entities don't just cost you visibility, they actively route customers to scammers. And half the risky domains were dead. Any domain you let lapse can be registered by someone else tomorrow, and the AI may still treat it as “official.”

Why this is a brand problem

The damage doesn't land on OpenAI. It lands on you: the customer loses money to a fake store and remembers your name. Trust is a brand's most valuable asset, and here it leaks through a channel you can't even see. In the middle of Black Friday season, when deal questions to AI assistants pile up, this is the wrong moment to park the issue.

What I'd do now

Publish one crawlable “official stores and retailers” page: every country store, every authorized retailer, and explicitly the countries with no store. Link it from the footer.
Run a domain inventory: which country-code and lookalike domains do you own, which ones lapsed? Don't just let expiring brand domains go.
Once a month, put the study's five questions to ChatGPT, Gemini and Perplexity and log the date and every domain. Country question first.

Bottom lineIf you don't tell the AI where your official store is, the counterfeiters will.

Got AI news? Submit news →