Misusing AI could compromise data security.

The revolution in artificial intelligence (AI) exploded from late 2022, especially after OpenAI launched ChatGPT. Soon after, major technology corporations also launched other AI tools – such as Microsoft's Copilot and Bing, and Google's Gemini – further increasing the usefulness of AI. These tools can assist with everything from writing code and analyzing data to synthesizing information in a short amount of time. This has led many users to become more interested in AI.

Using it while worrying at the same time.

Mr. Nguyen Hoai Sang, an office worker in Binh Thanh District, Ho Chi Minh City, said that since he started using AI software, preparing business reports, compiling information, and taking notes from internal meetings or client meetings via Google Meet and Zoom platforms has saved him about 40%-50% of the time compared to manual methods. However, after downloading the AI ​​tools, he noticed that his computer would freeze for 1-2 minutes, and its performance also decreased compared to before.

"Besides ChatGPT, Copilot, and Gemini, I searched the Google Chrome extension store and installed several AI tools like Sider, SearchGPT, and Tactiq. Since then, my computer has been slowing down. For example, typing in Word files is no longer smooth, and opening simple applications takes 5-7 seconds longer than usual, even though it's a good laptop on the market. As a result, my work is sometimes interrupted, causing distraction and frustration. But what worries me is whether this is due to a virus infection and AI infiltrating the device to steal data?" - Mr. Sang worried.

Meanwhile, Mr. Xuan Thanh, an IT employee in Ho Chi Minh City whose main job is to patch internal system errors within businesses, believes that AI greatly assists his work, requiring only descriptions and code writing to quickly resolve errors. However, he is also apprehensive about the security implications of AI.

"There are some tasks that require the immediate use of AI, so I'm forced to apply it. But I'm still concerned that the owners of some AI software might use the data collected during interactions and questions given to the AI ​​to infiltrate users' systems. Or, when the AI ​​collects data provided by users, malicious actors could ask about it, and would the AI ​​provide the information? Although there haven't been any incidents related to this yet, I'm still unsure about the safety of AI," Mr. Thanh said skeptically.

AI software can be the cause of computer malfunctions.

Be cautious when downloading AI.

According to Mr. Nguyen Huu Nguyen, Deputy Director of the Vietnam National Cyber ​​Security Emergency Response Center (VNCERT), computer lag and reduced performance after downloading AI software could be due to the complex features of AI software consuming a large amount of system resources, and not necessarily because of a virus.

However, according to Mr. Nguyen, users need to be cautious when downloading AI from unreliable sources, as it may contain malware to hack into the system. Users should also pay attention to data protection, carefully reading the privacy policy and terms of use of AI tools to understand how they collect and use data. In addition, access to the device should be controlled when using AI. Important documents should especially not be shared with AI to minimize risks.

Mr. Ta Cong Son, Head of the AI ​​Development Department - anti-fraud project, believes that computers slowing down or even freezing after loading AI could be due to insufficient RAM to handle its features. However, it is also a warning sign that users may be facing some risks.

"To ensure safety, users should restrict AI access. When an application requests permission, users need to carefully consider whether granting that permission is necessary and safe. A safe way to use AI tools is to use antivirus applications to scan and check. Additionally, users can use websites like Virustotal.com or Eset.com to check if the AI ​​contains malware or is safe," Mr. Son recommended.

At a recent high-level seminar on cybersecurity in 2024, Mr. Tran Son Binh, Business Director of AMD Vietnam, stated that users employing ChatGPT to write code, program, or summarize data could lead to system security issues. This is because the code provided by AI may contain malicious code, and the AI ​​can use this to exploit vulnerabilities and alter data structures, thus compromising the system's security and performance.

Mr. Binh emphasized: "Units need to control their personnel in applying AI to write reports and synthesize information based on user-provided data, especially content related to business and national secrets. Right now, we're still in a safe zone, but who knows if that data will remain secure in the future?"

AI will collect data.

According to Gemini, simply chatting with the app helps users improve Google's services, including the machine learning models Gemini relies on. In this improvement process, trained reviewers need to process user conversations. "Therefore, when using Gemini, do not enter any information you don't want the reviewers to see or that you don't want Google to use," Gemini advises.

Regarding Copilot, Microsoft states that it collects data from users through interactions. The AI ​​products it delivers are based on user input, and some are generated by the platform collecting data on other interactive activities.

This means that when users share any information with the AI, the application collects it to make the tool smarter. This implies that user data that they don't want to be disclosed to others may be visible to them.

Source: https://nld.com.vn/lam-dung-ai-coi-chung-du-lieu-kho-an-toan-196240330201849958.htm