ChatGPT could be sending shoppers to scam stores designed to steal credit card details – Automated Home

AI shopping tools can make online buying faster, but a new warning shows they can also send shoppers toward convincing scam stores.

Ask Silver found cases where ChatGPT surfaced fake retail websites that copied real brands, including Russell & Bromley and Dunelm. The risk is simple: shoppers may trust an AI recommendation, click a convincing store link, and enter payment details before realizing the site is not official.

What researchers discovered

The issue came to light after scam-checking service Ask Silver found examples of ChatGPT surfacing counterfeit retail websites in shopping-related results. Some of the sites copied real brands and used convincing layouts to appear legitimate.

One affected brand was Russell & Bromley, whose standalone website was no longer active after the company entered administration and was later bought by Next. Ask Silver found suspicious domains using variations of the Russell & Bromley name.

The issue was not limited to Russell & Bromley. Ask Silver also found cloned pages linked to active retailers, including Dunelm, showing that scammers can target both closed or changed brands and companies still trading normally.

How the fake stores operate

The fraudulent websites are designed to look authentic from the moment visitors arrive. They often feature polished layouts, product images, professional branding, and promotional banners that resemble legitimate e-commerce stores.

Many of the sites advertise dramatic discounts, sometimes claiming products are available at up to 80% below normal prices. Countdown timers and other urgency tactics are used to pressure visitors into making quick purchases.

Researchers documented one example where a fake Russell & Bromley website promoted products with discounts of up to 80%. The pricing appeared attractive enough to encourage purchases while remaining believable to shoppers unfamiliar with the retailer’s current status.

Why AI recommendations create a unique risk

Traditional search engines typically present multiple links, allowing users to compare sources before clicking. AI shopping tools work differently by compressing product discovery into a small number of recommendations that appear inside a conversational response.

That design creates an implied sense of trust. When an AI assistant presents a retailer as part of an answer, many users assume some level of verification has already occurred behind the scenes.

As a result, fewer people carefully examine web addresses before proceeding. Security researchers argue that this changes consumer behavior and creates a new attack surface that scammers can exploit through manipulated content and deceptive websites.

How scammers manipulate AI shopping results

Researchers say scammers can create cloned websites and misleading product pages that make fraudulent stores look more visible and convincing online. Those pages may then appear in AI-assisted shopping results when users ask for products or brands.

This is different from poisoning an AI model’s training data. In this case, the concern is that AI tools may surface bad web results when they search or summarize shopping information.

The result is still dangerous for consumers. A fake store can look credible in an AI-generated answer even when the website is built to collect payments or personal details.

Little-known fact: In one Ask Silver test, ChatGPT’s shopping results for Russell & Bromley included fraudulent source links alongside product recommendations.

Why Russell & Bromley became a target

The Russell & Bromley case highlights how changes in a company’s online presence can create opportunities for fraudsters. After the retailer entered administration, its standalone website was removed, creating confusion among consumers searching for the brand.

The business was later acquired by NEXT, but customers unfamiliar with the transition could still expect to find an independent Russell & Bromley website. That expectation made it easier for scammers to position fake domains as apparent successors.

Several suspicious websites used brand-related terms such as “official,” “London,” and “onlineuk” to increase credibility. To many consumers, those names looked plausible enough to pass a quick inspection.

Little-known fact: A spokesperson for ChatGPT said it had removed the fraudulent websites from its search index. Users of the AI tool can report sites that violate its policies through this form.

The payment traps shoppers should watch for

One of the most dangerous elements of these scams is the payment process. Fraudulent retailers often steer customers toward methods that provide little or no buyer protection if something goes wrong.

Security experts warn against websites that heavily promote wire transfers, cryptocurrency payments, or direct bank transactions. These methods can make recovering funds extremely difficult once a payment has been sent.

Legitimate retailers usually offer recognizable payment options and clear customer-service information. If a site pushes direct bank transfer, cryptocurrency, gift cards, or other hard-to-reverse payment methods, shoppers should treat it as a major warning sign.

What OpenAI has done so far

Following reports of the fraudulent recommendations, OpenAI stated that the identified scam websites had been removed from ChatGPT’s search index. The company also provides a reporting mechanism for users who encounter suspicious results.

Retailers affected by the issue have also taken action. NEXT acknowledged awareness of the fraudulent Russell & Bromley sites, while Dunelm encouraged customers to use only its official website when making purchases.

Although the immediate response addresses specific examples, experts argue that broader challenges remain. Verification systems for retailer identity and source authenticity are still developing as AI-powered shopping continues to expand.

A growing problem beyond retail

The retail examples fit a wider pattern of online scams becoming more convincing. Security firms and consumer-protection agencies have warned that criminals increasingly use cloned websites, phishing pages, fake ads, and impersonation tactics to trick shoppers.

AI can make that problem harder to spot because users may assume a shopping recommendation has already been checked. That makes source verification more important before entering payment details.

Little-known fact: More than half of victims reported losing money in 2026, more than double the 2025 level, with consumers ages 65 to 74 facing the highest monetary-loss rate.

What smart home users should know

Smart home shoppers may face similar risks because they often compare connected devices, subscriptions, and accessories online. If an AI assistant recommends a fake store for cameras, smart locks, sensors, or hubs, the financial and privacy risks can be serious.

A fraudulent smart home store could potentially advertise cameras, sensors, smart locks, or hubs at unusually low prices. Beyond financial loss, shoppers might also expose personal information that could later be used in phishing campaigns.

Consumers researching connected devices should verify listings through official manufacturer websites whenever possible. Cross-checking product availability directly with trusted retailers adds an extra layer of protection that AI recommendations alone cannot provide.

How to verify a retailer before buying

Experts recommend a simple verification process before entering payment information. Instead of clicking directly from an AI recommendation, users should manually type the retailer’s address or locate the official site independently.

Review platforms and consumer protection organizations can provide additional confirmation that a business is legitimate. Spending a few minutes validating a retailer often reduces the likelihood of falling victim to cloned websites.

TL;DR

• Researchers found cases where ChatGPT recommended counterfeit retail websites that appeared legitimate but were allegedly designed to collect payments and sensitive customer information.
• Scam operators reportedly use cloned websites, fake discounts, and manipulated online content to increase the likelihood that AI systems surface fraudulent stores.
• The danger is amplified because AI recommendations often carry implied trust, causing some users to skip basic checks they might perform on search engine results.
• Fraudulent retailers frequently encourage payment methods with limited buyer protection, making it harder for victims to recover funds after a transaction.
• OpenAI has removed reported scam websites from its search index, but experts say stronger retailer verification systems are still needed across AI shopping platforms.

This article was made with AI assistance and human editing.

If you liked this, you might also like: