ARTIFICIAL INTELLIGENCE AND CYBERSECURITY: THE CONTRIBUTION OF GPT AND BERT IN THREAT DETECTION AND INCIDENT ANALYSIS

Protecting information systems is becoming a growing challenge in the face of intensifying cyberattacks. Conventional detection solutions show certain limitations when it comes to anticipating and neutralizing sophisticated threats. This paper presents an innovative approach, leveraging advanced artificial intelligence models, BERT and GPT, to strengthen automatic threat identification and security incident analysis. Our approach is based on the combination of BERT to classify security alerts and GPT to generate detailed and enriched reports. Experiments carried out with the CIC-IDS2017 dataset revealed an impressive 97% accuracy in threat classification, with a significant reduction in false alerts. The results demonstrate that the combination of these models optimizes the analysis of security logs and allows for faster action against cyberattacks. This work illustrates the potential of technologies based on natural language processing to transform cybersecurity and pave the way for more autonomous and intelligent solutions.