• The CITIC researcher Rubén Pérez analysed the multiple practical applications of the deep compression capability online language holds.

A Coruña, March 24^th, 2026

The CITIC at the UDC hosted a new session of the TIC Talk breakfast series on Tuesday, March 24^th, entitled “Securing the Internet: When AI Learns the Language of the Network,” presented by researcher Rubén Pérez Jove.

During the talk, the way Artificial Intelligence, driven by Large Language Models (LLMs) and Transformers-based architecture, has opened the door to a new way of analysing network traffic was addressed. In this regard, Rubén Pérez spoke about the possibility of applying these models so that AI can read and understand the data packets circulating on the Internet, in the same way that it processes natural language.

The session centred on the so-called foundational models of network traffic, delving into their operation and the methods used to teach them to interpret the structure and semantics of this data. This advanced comprehension capability was presented as a key element for multiple applications, especially in the field of cybersecurity.

In this context, it was highlighted how these models are revolutionizing critical tasks such as anomaly detection, intrusion prevention, asset identification, and traffic classification. “By understanding the network’s common ‘language,’ AI is proving capable of identifying malicious or unusual behaviour with greater effectiveness than traditional systems,” the researcher stated.

The presentation was given by Rubén Pérez, an FPU predoctoral researcher at CITIC, under the supervision of José Vázquez Naya and Alejandro Pazos. The content presented is based on an exhaustive systematic review conducted as part of his doctoral thesis.