There are many ways to train artificial intelligence, but some are more controversial than others. A South Korean study created DarkBERT, an AI trained using data collected from the Dark Web. In this article, we explored the benefits of this approach, the challenges faced in training it, and the potential applications of DarkBERT in the fight against cybercriminals.
DarkBERT is created using data collected from the Dark Web
Artificial intelligences such as ChatGPT are trained from a large, transparent, public database. But in the case of DarkBERT, things are different. The researchers had to connect directly to the Tor network to access data from the Dark Web. They then began to create a pure database of information collected on the dark side of the Internet.
DarkBERT was designed to study cybercriminals in depth to better combat them. According to the research team, DarkBERT was able to create a better picture of the Dark Web than other similar artificial intelligences, including the RoBERTa model developed by Facebook in 2019.
DarkBERT, an AI to fight cybercriminals
But how can DarkBERT be used to fight cybercriminals? The technology has several potential applications in this area. It can detect websites dedicated to selling ransomware or sensitive data. DarkBERT can also monitor Dark Web forums for illicit information exchanges.
The existence of DarkBERT shows that fighting cybercrime can take many forms and that artificial intelligence can be a very reliable tool in this fight. While this approach raises questions about privacy, it also offers great potential for online security.
