Redazione RHC : 17 September 2025 18:44
The Artificial Intelligence Security Governance Forum was held in Kunming, Yunnan Province, southwestern China, as part of the National Cybersecurity Publicity Week 2025.
The event offered an opportunity to discuss the risks and challenges associated with AI, governance measures, and developments related to the security of applications and algorithms. Researchers, technicians, and representatives from various sectors shared experiences and practical results, highlighting how AI is revolutionizing numerous fields, including cybersecurity.
At the security fair, technicians emphasized how large companies are now exposed to a wide range of cyber attacks. The threats range from more common, such as web application attacks, to more sophisticated and difficult-to-detect forms, such as Trojan communications that encrypt entire channels before attacking.
Technician Zou Hong, speaking during the presentation, explained that in the past, cyber defense was based mainly on manual interventions, with long and costly response times. The introduction of artificial intelligence has instead allowed for faster and more efficient analysis systems, thanks to models capable of processing and dividing attack alerts into multiple layers.
Zou illustrated a hybrid architecture that combines models of different sizes to perform a three-layer screening, capable of reducing alert noise with an effectiveness of over 98%. This approach allows us to distinguish truly dangerous attacks from irrelevant ones, focusing efforts on the most concrete threats.
For cases that still require manual analysis, large-scale artificial intelligence can make a decisive contribution by supporting the identification and tracking of attack sources. This further accelerates defense and response processes.
Once the data has been collected and analyzed, technicians prepare detailed reports that are sent to the relevant law enforcement agencies and national departments. This ensures institutional coordination and structured management of cyber threats.
Finally, experts have drawn attention to the misuse of AI by criminals, who exploit large models to launch attacks. The answer, according to Zou Hong, is to rely on a mirror approach: using artificial intelligence to fight artificial intelligence.
By training models to recognize attack frequencies and patterns, security devices can be mobilized to automatically intercept them and ensure a more robust defense.
Redazione