Within a pas few years, technology-based scams have become one of the most pressing issues in Vietnam. According to the Ministry of Public Security, more than 20,000 phone and online scam cases were reported in 2024 — a 65% increase from the previous year, causing losses estimated at 1.89 trillion VND (approximately 75.6 million USD).
These figures are not just numbers. They represent growing concerns about digital literacy and the urgent solutions to protect users from increasingly sophisticated fraud schemes.
lost to online scams
phone users have been scammed
reports of spam or scam calls
of users have reported fraudulent calls
of users have reported unwanted calls
(Source: Ministry of Science and Technology, Vietnam, 2024)
As technology continues to shape every corner of our lives, Artificial Intelligence (AI) is advancing faster than ever but so are digital threats. Scammers are adapting to online platforms to trick users in real time.
When deception can unfold within seconds of a phone call, how can users tell what's real and what's a trap? — This question became the foundation of our project.
We wanted to explore how technology could go beyond being just "smart" to become safe. Our solution is a mobile application powered by AI that detects suspicious patterns during live phone calls and provides instant alerts to help users stay one step ahead of scammers.
During our research, we discovered that scam calls and messages often manipulate human emotions rather than technical tactics. Scammers exploit psychological triggers such as greed, fear, and empathy to trap victims.
Messages or websites promising rewards, Telegram “investment” groups, or fake job offers with unusually high salaries.
Scammers impersonate police officers or government officials, threatening legal action and demanding urgent money transfers to a “safe account.”
Messages on Messenger or Zalo pretending to be relatives asking for money due to emergencies or accidents.
Scammers carefully prey on emotions to push victims into quick, irrational decisions. By using persuasive language and urgent storytelling, they create a false sense of trust and control the victim’s emotional state.
To avoid harm, it’s crucial to disrupt at least one step in the victim’s psychological spiral. Humans often recognize the danger too late, but AI can detect it early — right as the call is happening.
Our app is designed to do exactly that: intervene at the right moment to block scam behavior before emotion overrides logic.
How might we block scam calls before they reach the user?
How might we deliver continuous real-time scam alerts during the conversation?
How might we immediately comfort and guide victims?
Warning Screen
1. Automatic Number Verification: The system automatically checks the caller’s identity by comparing it against the National Verified Number Database.
2. Early Warning System: If the incoming call meets any of the following conditions, an alert instantly appears on the screen:
• The number is not verified in the national database.
• The number’s data does not match existing verified records.
This helps users quickly recognize potential risks and decide whether to answer or ignore the call.
1. Real-time AI Conversation Analysis: An AI Detection engine runs continuously during the call, analyzing both voice tone and transcribed text (Speech-to-Text). It interprets language meaning, sentiment, and emotional cues to detect suspicious patterns.
2. Risk Scoring System: The system assigns a risk level score in real time, displayed visually through color and vibration feedback:
•Safe (0–20%): No suspicious activity detected.
•Caution (30–60%): Signs of potential manipulation — device vibrates and shows a yellow warning screen.
•High Risk (70–100%): Strong evidence of scam behavior — red screen appears with urgent vibration and alert signals.
AI Analysis Screen
1. AI-powered Call Analysis: The post-call screen provides a comprehensive summary that helps users recognize potential threats and take immediate action.
• Information Summary: The system records and analyzes all call data in real time.
• Risk Assessment: AI calculates a risk level score (e.g., potential scam) based on voice tone and conversation content.
• Guidance & Protection: The system instantly provides safety recommendations and verified next steps to help users protect their personal and financial information.
2. Report Suspicious Numbers: Users can submit details of the suspicious call — including the phone number, time, and AI analysis result — to the National Database and community reporting system.
This collective reporting network helps expand national scam prevention efforts, protecting millions of users from similar fraudulent tactics.
ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
0123456789
!@#$%^&*()
ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
0123456789
!@#$%^&*()
