Hieu PC warns that the habit of using AI can cause Vietnamese people to reveal confidential data

(Dan Tri) - Carelessly uploading personal images, identification documents or internal documents to AI tools can put users at risk of data leaks, becoming targets of cybercriminals.

On October 11, in Hanoi, within the framework of the conference "Information Security in the AI ​​Era" organized by the FSB Institute of Management & Technology (FPT University), cybersecurity expert Ngo Minh Hieu (Hieu PC) gave many warnings about cyber attack methods with the support of artificial intelligence.

Carelessly posting data to AI, risks revealing sensitive information

According to expert Ngo Minh Hieu (Hieu PC), one of the current worrying risks comes from the uncontrolled habit of using AI.

Many people are willing to upload personal photos, identification documents, contracts, financial documents, and even internal business information to AI chatbots for analysis, summary, or editing.

Experts warn that data provided to some AI services may be stored, processed or used against users' wishes. In particular, platforms of unknown origin pose a potential risk of information leakage if the security system is compromised.

Cyber ​​security expert Ngo Minh Hieu (Hieu PC) warns that fraud tricks are becoming increasingly sophisticated thanks to the support of AI (Photo: FPT University).

Therefore, businesses need to clearly stipulate which AI tools are allowed to be used, and absolutely do not arbitrarily post confidential documents on public platforms.

Not only documents, images and videos posted publicly on social networks can also become a data source for criminals to exploit.

According to this expert, people who frequently appear on the Internet, especially business leaders or people with the right to approve financial transactions, are at risk of having their identities faked using Deepfake and DeepVoice.

Images and voices shared publicly on social networks can be exploited by criminals to create fake content using AI (Illustrative photo: Tuan Anh).

Experts cited a scam in Hong Kong (China) in 2024, when a financial employee participated in a video conference with supposed leaders and colleagues.

In fact, the characters in the meeting were all faked using Deepfake. Trusting instructions from superiors, the victim transferred a total of about 25 million USD to the crooks.

The incident shows that image and voice are no longer a reliable enough basis to verify identity in the digital environment.

AI is helping cybercriminals become more sophisticated in their fraud

According to experts, cyber criminals are increasingly operating in an organized manner, divided into specialized departments similar to businesses.

Previously, phishing messages often had spelling errors and unnatural expressions. Currently, AI can create smooth scenarios that strike victims' psychology.

Real-time translation technology also helps subjects overcome language barriers, while AI chatbots can automatically approach and screen people who are likely to fall into traps before transferring them to real people to commit fraud.

The habit of uploading personal documents, contracts and internal documents to AI tools can pose a potential risk of leaking sensitive information (Illustration: Minh Trang).

Citing data at the conference, experts said that in 2024, Vietnam will record more than 6,000 online fraud cases, causing losses of over 12,000 billion VND.

Another tactic that is warned about is spreading malicious code under the guise of familiar public service, banking or utility applications.

Fraudsters lure users into installing APK files from unofficial sources, then request Accessibility permissions on Android phones to take control of the device.

Experts say two places crooks often look for are note-taking applications and photo libraries, where many people save passwords, PIN codes, identification documents or bank account information.

AI is helping cybercriminals create sophisticated phishing scenarios, making it difficult for users to recognize with common signs (Photo: Viet Anh).

To prevent this, experts recommend that users turn on two-factor authentication, do not install applications from unfamiliar links, and limit the publication of sensitive data.

In particular, before calls requesting urgent money transfer, it is necessary to verify through an independent communication channel, even if the image and voice resemble an acquaintance.

At the enterprise level, Hieu PC proposes a 4-layer security model including people, process, technology and AI management. In particular, important tasks such as transferring money, deleting data or making system changes require human verification and approval mechanisms.

Dr. Nguyen Van Duc, representative of the FSB Institute of Technology Management, said that the institute has a Master of Information Technology program oriented on Artificial Intelligence of Things (MIT-AIoT). The program aims to train the capacity to design, integrate and operate smart systems instead of just focusing on each individual technology.

Dr. Nguyen Van Duc, representative of FSB Institute of Management and Technology Photo: FPT University).

AIoT is a combination of AI and Internet of Things (IoT), applied in production, transportation, urban management and business operations.

However, as the number of connected devices increases, the risk of cyber attacks also becomes more complex, creating a need for human resources knowledgeable in both AI, connection infrastructure and data security.