Chinese enterprises fight AI hackers with robotic guardians

BEIJING -- As the Chinese adage goes: "apply the same tactics to the person who used them on you." At a time when artificial intelligence increasingly looms behind cyberattacks, Chinese enterprises have started to employ generative AI technologies to build up defense against such AI threats.

Qi-Anxin Group, a leading Chinese cybersecurity company, released its large-scale model QAX-GPT in March, expecting it to assist in the development of security products, detection of threats and areas of vulnerability, and analysis of internet-related crimes.

This model boasts investigation and judgment abilities close to that of an intermediate-level security expert, while its efficiency in terms of alarms and judgments is more than 60 times that of manual efforts, said Qi Xiangdong, chairman of Qi-Anxin.

Since ChatGPT, developed by OpenAI, took the world by storm in late 2022, tech giants and startups worldwide have been rushing to join the AI race by launching similar AI chatbots, as well as industrial applications based on large language models.

Experts believe that generative AI tools lower the barrier to entry for threat actors with limited programming abilities or technical skills and warn that hackers have started using generative AI tools to create malware, dark web and other tools to carry out cyberattacks.

Such AI tools, however, are a double-edged sword for network security. There are also cases where generative AI tools have been fashioned into a weapon to counter cyberattacks.

A lack of network security personnel and resources, leading to alarms being neglected or mishandled, is the biggest vulnerability facing network security, according to a survey by Qi-Anxin.

"To avoid affecting operations, 99 percent of alarms suggesting cybersecurity threats require expert analysis. However, the number of experts in any company is limited compared to the tremendous number of alarm instances. The comprehensive analysis system empowered by AI will therefore greatly boost security defense," said Qi.

Compared with a human expert, QAX-GPT learns much faster, gaining its expertise from a corpus of analysis, reports and articles related to cybersecurity, said Zhang Zhuo, vice president of Qi-Anxin.

"Challenges facing cybersecurity are scarcity of experts, human alarm fatigue and efficiency bottlenecks. The large-scale model enables us to transform from reproducing goods to reproducing expert experience," said Zhang.

China has been a major victim of hacking and cyberattacks. More than 42 million malware attacks were detected in China in 2020, according to a report released by the cybersecurity watchdog, the National Computer Network Emergency Response Technical Team/Coordination Center of China.