A group of four students from Swinburne University Vietnam have successfully completed research and testing of an artificial intelligence (AI) tool capable of identifying and filtering malicious and fake websites in just seconds.
From idea to action
Le Nguyen Viet Cuong, the student group leader, said that the group members have noticed the recent emergence of many scam websites with increasingly sophisticated methods, while current solutions relying on human intervention are insufficient to prevent them.
Therefore, the group of four students created the website https://ai.chongluadao.vn/, which can directly analyze the website the user is viewing to assess its safety. "By applying AI and analysis, the group can identify scam methods and provide information to users to raise their awareness," Cường explained.
According to Truong Duc Sang, a member of the team, the group used popular AI models readily available on the market such as ChatGPT, Gemini, Claude, Grok, Deepseek, etc. After analyzing and comparing the results of the models, operating costs, and accuracy assessments, the group selected the best and most suitable application for the project.
A group of four students at Swinburne University Vietnam are researching an artificial intelligence application capable of identifying and filtering malicious websites in just seconds. (Photo: DUY PHÚ)
Another member, student Nguyen Van Huy Quang, pointed out that the applications currently on the market that filter malicious and fraudulent websites use information already collected by the provider in their database, not content directly from the website the user is viewing. "Using AI analysis can help users identify malicious websites and understand how dangerous they are in real time," Quang explained.
Members of the team promoting this application can easily access it via the website, requiring only an internet connection. Users can provide the URL of a website suspected of being fraudulent, and the AI can then analyze and verify it immediately.
Community project
Young students studying AI and software believe that for successful project implementation, knowledge learned in school is not enough; what is most important is determination, self-learning, and a thirst for knowledge.
Group member, student Nguyen Vinh Khang, expressed that this is a community-oriented project, and the group hopes to widely disseminate information to users, helping to raise awareness in the community about the risks of encountering fake websites, malware infections, and being scammed out of data or assets. At the same time, the group also hopes to receive support from relevant authorities to promptly prevent fraudulent websites from spreading.
Elaborating on the implementation process, Duc Sang said the biggest challenge was that the project used AI models, large-scale language models that had only been around for 1-2 years. "With previous projects and those who came before, the obstacle was the very limited documentation. The team had to painstakingly research, experiment, and try many times, and not every attempt yielded the expected results, so it was quite time-consuming," Duc Sang recalled.
According to team leader Le Nguyen Viet Cuong, the most difficult part was data collection, because a lot of data is needed to train the AI, while the information and data on the internet are only relatively accurate.
"Because we knew there was no perfect data source, the team members had to manually input and edit the data. After entering the data, waiting for the AI to return the results was both nerve-wracking and exciting. Large datasets could take an entire day to process, and after the results came back, we continued testing. There were times when we thought we had succeeded, but the results showed low accuracy. Although disappointed, the whole team was determined to find a solution," Cường recounted.
Accuracy up to 98%
Cybersecurity expert Ngo Minh Hieu, Director of the Chongluadao.vn project, said that in late 2024, while teaching cybersecurity at Swinburne University Vietnam, he discovered a group of students who were very enthusiastic and quick to apply AI. He then gathered them together to discuss their long-held idea. The malicious website filtering tool, developed by the student group using the knowledge and experience of the Chongluadao.vn project members, has been completed and achieves an accuracy rate of over 98%.
The project's "patron" commented that the tool helps detect phishing websites very early, even before a database is established. In addition, the tool has a user-friendly interface and can be integrated into the interfaces of many different parties.
"We hope this product can be used by people, businesses, and large corporations. We also hope that the students in the research group will become the F2 generation of the Chongluadao.vn project," Mr. Hieu expressed.
Source: https://nld.com.vn/ung-dung-ai-de-loc-website-lua-dao-196250422215400349.htm
Comment (0)