Generative artificial intelligence tools and search assistants like ChatGPT are increasingly directing users to fraudulent websites and malicious online schemes, according to new research analysing web traffic and AI search query outputs.
The findings highlight a growing vulnerability in how large language models crawl, index, and surface web links.
Unlike traditional search engines, which rely heavily on domain authority and established indexing algorithms to filter out malicious content, AI search assistants summarise vast amounts of web data, occasionally scraping and prioritising low-quality or compromised sources that contain phishing links or fraudulent digital stores.
New findings from Eset, Europe’s largest cybersecurity company, have raised concerns about users who use AI tools to search for products, software and information online, assuming links provided in an AI response have already been checked for safety. In fact, using ChatGPT’s search results has been reported to direct users to high-risk websites already identified as unsafe.
The sites identified included fake online stores, cryptocurrency scams and imitation login pages, as well as AI-generated results linking to malicious code that may steal users’ information.
“The risk is that we start placing blind faith in the answers these tools provide. Someone asks an AI tool for help, follows the link it provides and ends up on a fake login page, scam website or malicious download,” says Scott Leman, Eset NZ country manager.
Because that recommendation has come from a tool they trust, they may be less likely to question whether the destination is genuine or safe, he says.
“That misplaced trust creates a significant entry point for malicious actors who optimise fraudulent sites to be picked up by AI scrapers.”
How scammers exploit AI models
To manipulate crawling data, attackers publish high volumes of keyword-optimised, AI-generated content designed to rank quickly and get picked up during automated web-scraping cycles.
Then it redirects users to malicious landing pages or domains by following a link, opening a document, scanning a QR code or entering sensitive information into a fake website, within AI answer summaries as legitimate recommendations for customer support, online booking, or retail platforms.
Scammers also embed fraudulent telephone numbers and external links into legitimate-looking forum posts or press releases, which AI tools aggregate and present as factual solutions.
“For years businesses could reasonably assume that when a new security weakness was discovered there would be some time to understand it, fix it and protect themselves,” Leman says.
“That breathing space is disappearing. AI can now analyse a newly released security patch, work backwards to identify the flaw it was designed to fix and help turn that information into a working attack extremely quickly.”
For SMEs, Leman says the growing threat environment is particularly challenging, as they often lack dedicated security teams that can continuously monitor suspicious activities.
Businesses are warned not to rely solely on traditional SEO or passive domain management to safeguard their market footprint.
“For smaller businesses, continuous monitoring and rapid response are becoming increasingly important, particularly as the window to detect and contain an attack gets shorter.
“As attacks accelerate, the ability to detect and respond is becoming just as important as trying to prevent them in the first place.
“Antivirus remains an essential first line of defence, but not every modern cyber attack involves a virus. There may be no virus file for antivirus software to find because the criminal is effectively walking through the front door using the user’s own key.” Leman adds.
