Phishing techniques have “evolved” and continue to “exploit the most vulnerable element in security systems: human psychology.” This was the starting point for the development of Check Point’s new technology, DeepBrand Clustering.
The feature also follows the brand’s anti-impersonation technology, a technology created by the brand in 2023 that uses “artificial intelligence, natural language processing, image processing, and reasoning” to “detect and prevent attempts to impersonate a brand’s identity and match URLs and websites to established brand pages.”
DeepBrand Clustering is now being touted as “the next evolution in brand counterfeiting prevention,” with a process divided into two stages, “learning and criminalization.” In the first, a “neural network” is built using “features extracted from monitored web pages, which originate from Check Point’s global traffic.”
All of these attributes (text, graphics, and favicons, i.e. everything that helps determine the legitimacy of a brand) are then analyzed and “consolidated” by DeepBrand Clustering’s AI engine, which creates groups (clusters) containing data that identify brands.
The next step is the criminalization phase. Here, “the inference process determines whether the examined web page belongs to any of the existing groups,” which will serve to assess “potential malicious attempts at trademark counterfeiting.”
To date, DeepBrand Clustering has “cataloged over four thousand different brands” — of which “over two hundred have been counterfeited in over four thousand malicious attacks.”
