dark side of the internet

Darknet is a secret network where connections are established only between trusted people and devices. At the same time, special software solutions, non-standard protocols and ports, anonymizers and other technologies are used to ensure confidentiality. An anonymous network is a system of unrelated virtual tunnels and servers that provide encrypted data transmission. The darknet is often used by criminals: arms trade, drugs, stolen data, malware and even a murderer can be found here.

A team of South Korean researchers turned their attention to this data source to train their artificial intelligence, called DarkBERT, which indexes some of the most secret areas on the internet. Researchers say their technology has noble purposes: it is seeks to shed light on new ways to combat cybercrime, where natural language processing is increasingly used.

As detailed in an article titled “DarkBERT: A Language Model for the Dark Side of the Internet”, which has not yet peer-reviewed, the team connected their model to the Tor network, a system for accessing the dark web. After that, he set to work creating a database from the raw data he found.

The team says their new LLM (language model) is much better at understanding the dark side of the Internet than other models trained to perform similar tasks, including the RoBERTa developed by Facebook researchers in 2019.

Our evaluation results show that the DarkBERT-based classification model outperforms known pre-trained language models.
– write the authors of the article.

The team speculates that DarkBERT could be used for a variety of cybersecurity tasks, such as detecting sites that sell ransomware or trade sensitive data. It can also be used to scan the numerous forums on the dark web, updated daily, and monitor them for illegal information exchange.