How neural networks help fraudsters empty the wallets of Russians: the new "data poisoning" scheme
Trust in artificial intelligence recommendations when choosing products is becoming a dangerous trap. Fraudsters are actively exploiting vulnerabilities in search neural networks to redirect users to phishing resources designed to steal personal data and money. This scheme, which experts call "data poisoning," has already led to financial losses among trusting Russians.
How the "poisoning" mechanism works
Analysis shows that attackers are deliberately "clogging" the internet with fake content. They generate fake reviews, automated forum posts, and false data for indexing. This entire information attack is aimed at distorting how algorithms perceive and rank sources.
As a result, the neural network mistakenly identifies a fraudulent site as a reliable seller. Most often, clothing and furniture stores are targeted, masquerading as recently closed brands. Criminals copy brand logos, images, and product descriptions, and to attract victims, they promise discounts of up to 80%. Chatbots integrated into search engines, in some cases, directly send buyers to these fake resources.
How to protect yourself from AI fraud
The main protective tool is cross-verification. Before making a purchase based on a chatbot's advice, it is necessary to compare the obtained data on other search platforms and find the official original source of information. At the time of writing this analysis, the identified fraudulent resources have already been removed from the search index, but this does not guarantee that new ones will not appear.
It is important to remember the limitations of the technology. AI algorithms often provide outdated or irrelevant data. Therefore, it is wiser to perceive a neural network's recommendation as a starting point for your own research, rather than as a ready-made guide to action.
Expert opinion: This scheme is a vivid example of how attackers adapt to new technologies. While search engine developers fight "data poisoning" after the fact, users must develop critical thinking and not blindly trust AI recommendations. In the era of generative AI, where content is created automatically, trust must be backed by verification.