AI Security: A Double-Edged Sword

Artificial Intelligence (AI) is reshaping every aspect of life, and cybersecurity is no exception. However, the rapid development of this technology is posing a big paradox: AI is both the most solid defense shield and the most dangerous attack weapon. At the BSides Hanoi 2025 event with the theme "AI Hack You", leading experts came together to "dissect" the nature of this double-edged sword, sketching a panoramic picture of cyber warfare in the new era.

According to Mr. Vu Duy Hien, Deputy Secretary General and Chief of Office, National Cyber Security Association, AI is currently one of the technologies that plays a particularly important role, supporting individuals and businesses in many fields and helping to process work faster and more effectively. However, in the cyber security space, AI is a double-edged sword.

AI - Smart 'bodyguard'

There is no denying that AI is making great strides in the cybersecurity industry. “On the positive side, AI helps us detect attacks early, analyze data, and strengthen network defenses. Many smart security tools created thanks to AI have significantly reduced response time to threats,” Mr. Vu Duy Hien affirmed.

Mr. Vu Duy Hien - Deputy Secretary General and Chief of Office, National Cyber Security Association - at the Workshop.

This power is clearly demonstrated through the sharing of Ms. Huynh Ngoc Khanh Minh, a member of the non-profit network Anti-Phishing. She said that in the past, when anti-phishing work mainly relied on manual processes, experts had to verify each suspicious link one by one, causing a huge delay in the tens of thousands of reports per day.

To overcome that limitation, the Anti-Fraud team has developed a new AI tool. “Users just need to paste the link, AI will automatically check and display the results in 30 seconds to 1 minute with an accuracy of up to 98% ,” said Ms. Khanh Minh. “ Thanks to that, the verification time is shortened dozens of times compared to before, helping people proactively detect and prevent the risk of online fraud effectively and completely free of charge.”

Globally , leading cybersecurity platforms have also deeply integrated AI. The Darktrace system uses self-learning AI to build an "immune system" for businesses, automatically detecting the smallest anomalies that could be signs of an unknown attack. Similarly, the CrowdStrike Falcon platform uses machine learning and behavioral analysis to prevent sophisticated attacks, instead of relying solely on known malware "signatures".

These tools show how AI is helping cybersecurity professionals move from reactive to proactive, stopping threats before they can cause damage.

AI - A weapon in the hands of the bad guys

The other side of the coin is equally brilliant. It is the superior capabilities of AI that have turned it into a coveted tool for cybercriminals. Mr. Vu Duy Hien pointed out: “On the other hand, cybercriminals also use AI to develop more sophisticated forms of attack, capable of overcoming many layers of traditional defense. The speed, scale and accuracy of cyberattacks have therefore increased dramatically.”

Deeply Personalized Phishing Attacks: With large language models (LLMs), hackers can create a series of emails and messages with natural language and context appropriate for each target, making even the most cautious users fall for them.

The Rise of Deepfake: Cybercriminals can fake a CEO's voice or video to request urgent money transfers. One famous case involved criminals using AI to fake the voice of a British energy company's CEO to successfully defraud $243,000.

Polymorphic Malware: AI can be used to create malware that can automatically change its source code after each infection, rendering traditional antivirus software ineffective.

Automating the attack process: Hackers can deploy AI to automatically scan millions of systems for vulnerabilities, carry out distributed denial of service (DDoS) attacks, or crack passwords at breakneck speeds, amplifying the scale of the attack many times over.

The human factor and the never-ending race

Faced with this reality, experts agree that the cybersecurity war in the AI era is a race between "good AI" and "bad AI". And at the heart of this race, the human factor becomes even more important.

Mr. Truong Duc Luong - Chairman of VSEC - at the Workshop.

Mr. Truong Duc Luong, Chairman of Vietnam Cyber Security Joint Stock Company (VSEC), said that with 22 years of experience, he realized the lack of opportunities for collaboration and sharing of in-depth knowledge between the Vietnamese cybersecurity team and the world, while the international security community is very active. This poses an urgent need to build a strong community.

Responding to this concern, Mr. Vu Duy Hien said: "AI is an irreversible trend, so instead of worrying, the cybersecurity community needs to cooperate, share experiences and together build a safe defense ecosystem that adapts to the new era."

In addition, the challenge is not only on the technical side. According to lawyer Nhi Pham, although Vietnam has legal documents, investigation and handling still face many challenges because violations often cross-border, making it difficult to trace and recover assets. According to her, to minimize risks, it is essential to have close cooperation and mutual understanding between the technical and legal teams.

Ultimately, the key to victory is to continuously improve your capabilities. Mr. Hien emphasized that preventing cybercrime in the AI era requires every user and every business to constantly update technology, and in particular, cybersecurity experts need to improve their professional capabilities and develop new, proactive solutions.

The double-edged sword of AI has posed a huge challenge but also an unprecedented opportunity for the cybersecurity industry. The future of cyberspace depends on how we use and control this powerful tool. The confrontation between defensive AI and offensive AI will be an endless arms race, and victory will only belong to the side that is better prepared, cooperates more closely, and constantly learns and innovates.

Huy Quang

Source: https://vtcnews.vn/bao-mat-bang-ai-con-dao-hai-luoi-ar970604.html