04 Aug Detecting the most difficult phishings… is possible
Detecting phishings in languages that have special characters, like Japanese, Hindu, Arab, Russian or Greek is not so easy. But it is not impossible with PhishingHunters solutions like Intel Modules.
One of the largest phishing databases in the world
First step to see if there is a phishing in a domain is identifying it, through Machine Learning. When we search a phishing in English, for example, we use an algorithm to recognize the most used images in the web, like the logo. This way we can detect if these images are illegal or fraudulent respecting the brand. Second step is extracting the texts and introducing them into an algorithm to see if they are natural language. When we finish that, we give the website a fraud scoring.
What is the difference between this process and the one with special characters in the language? That, sometimes, images are mixed with text. So, we have to use an OCR (Optical Character Recognition) that allows us to extract the text from the images. Once it is separated, the text has the same process as it was in English (indeed, it is always translated to that language).
Once a phishing is found, it is incorporated to our database. In PhishingHunters we have one of the biggest live feed databases in the world, with more than 2.000.000 registered phishing attacks. And it continuously grows.
This process is possible because we have developed our own system of detection and closure of phishings. Based on powerful tools of Machine Learning that are able to detect illegitimate or phishing domains through a proactive and 24 X 7 monitoring of the networks. This monitoring includes a constant feeding of filters and keywords. This way, the system can recognize phishing at an early stage and add them to the database.
Once the phishing is detected, we inform each client through our Social Media & Intelligence department.
Do you want to know how we work to keep your business safe of attacks?