(Central News Agency reporter Guo Fangjun, Zurich, 17th) Gioia da Silva, a technology columnist for the New Zurich Zeitung (NZZ), wrote an article stating that in July this year, autonomously operating AI agents carried out an unprecedented attack on Taiwan, which was quite sophisticated, and there were signs that China was involved in this operation. At the end of the article, it is emphasized that solid national defense must strengthen digital defense capabilities and protect digital sovereignty.
An editorial of Neue Zurich Zeitung (NZZ) on the 15th pointed out that Taiwan suffered an unprecedented wave of cyber attacks in July this year. The attackers used AI proxy tools with the ability to act autonomously to steal files, invade email services, and create backdoors in government network applications. This incident shows that AI may have evolved from an auxiliary hacking tool to an attack force that can independently find and exploit vulnerabilities.
The editorial explains that in the early stages of the attack, humans still set targets and monitor actions, but in the subsequent process, most of the attack work is completed by AI agent tools. The attack system can take multiple paths at the same time, continuously evaluate the probability of success, and divert resources to more feasible attack methods; if one of the paths is blocked, other agents can be sent to find new breakthroughs, and the strategy can be continuously adjusted in the process.
The article also stated that the outside world is still unable to confirm the identity of the attacker behind the scenes, but there are signs that China may be involved, which has also triggered high alertness in Taiwan's security community about AI changing the network's offensive and defensive structure. Compared with traditional hacker attacks, the biggest threat of AI agent tools is to reduce attack costs and increase speed and scale. Today's AI attack capabilities are as effective as a team of professional hackers in the past. However, although defenders also use AI to find system vulnerabilities, fixing them still requires time and professional manpower.
In addition, there is a high degree of asymmetry in network attack and defense. Attackers only need to find a vulnerability, but defenders must try to block all possible attack paths. The development of AI is accelerating and magnifying this gap, making attacks more frequent and more difficult to predict.
The article also emphasizes that signs that AI agent tools may break through existing security regulations have long attracted attention. If AI with autonomous decision-making capabilities is weaponized by a malicious country or criminal organization, whether it is a startup, a large enterprise, a city or a country, it may be attacked.
At the end of the article, it is pointed out that the situation Taiwan is facing now will be a challenge that Western democracies must face in the future. In addition to strengthening government information systems, countries should also improve the digital resilience of critical infrastructure and re-examine their national security architecture. National security not only involves military and physical defense, digital defense capabilities are also related to national sovereignty. Today, if a country does not protect its digital sovereignty, it may also lose the ability to maintain national autonomy.
