Identifying the Challenge
The Social Problem:
ScamSniper tackles the rising crisis of online scams in Asia, where digital fraud is causing
widespread financial and social harm. In Singapore, scam cases surged from 5,300 in 2016
to over 46,000 in 2023, with victims losing over S$650 million. Malaysia reported RM1.3
billion in cybercrime losses between 2021 and 2023 , while Vietnam received over 14,000
scam-related complaints in 2023. These scams ranging from phishing and fake job offers to
impersonation and investment fraud which preys on the trust of everyday people, leading
to lost life savings, emotional trauma, and fractured families. Beyond individuals, the
region faces a darker trend: in 2025, over 7,000 people were rescued from scam syndicates
in Myanmar after being trafficked and forced to run fraud operations. The scale and
sophistication of these operations demand smarter, tech-driven solutions. ScamSniper
responds by using AI to detect scams in real time, empowering users before irreversible
harm occurs.
Innovation and Uniqueness
Why Our Project Stands Out:
Our project, ScamSniper, offers an intelligent and scalable solution to the fast-evolving
scam landscape in Asia. VerifyAI is a real-time AI scam detection tool that analyzes
suspicious images such as phishing messages, love scams and fake job offers using
machine learning and natural language processing. What makes ScamSniper truly
innovative is its self-improving AI pipeline. As users upload new scam content, the AI
continuously learns from these real-world examples, staying up to date with the latest
scam tactics and linguistic patterns. This adaptability gives our system an edge that
traditional manual reporting and static rule-based systems cannot match. By enabling the
AI to evolve faster than scammers can pivot, ScamSniper creates a feedback loop that
actively cripples scam operations by detecting fraud before it spreads, and making it
increasingly difficult for scammers to succeed. Combined with a user-driven forum,
ScamSniper empowers people while making the scam industry unsustainable through
speed, accuracy, and scale.
Insights and Development
Learning Journey:
During the development of ScamSniper, we gained a deep understanding of how AI can be
leveraged for real-time fraud detection and prevention. A major insight was realizing how
rapidly scam tactics evolve, and the importance of building an adaptive AI system that
continuously learns from new data. One challenge we faced was training the AI to
differentiate between subtle scam patterns and legitimate content, which required us to
fine-tune our model for accuracy and reduce false positives. Another challenge was
integrating the community-driven forum with AI-powered detection, ensuring that user
input was seamlessly incorporated into the system. Overall, the project taught us the
critical balance between innovation, scalability, and user engagement in creating a
solution that is both effective and sustainable.
Development Process:
The development of ScamSniper involved prompt engineering to evaluate scams based
on a predefined set of criteria. The challenge was to craft effective prompts that could
guide the model in detecting subtle scam patterns in images and text. We created a
structured set of evaluative criteria, such as tone, language patterns, and red flags like
urgency or impersonation to assess whether a message was likely to be a scam. This
approach allowed for flexible, real-time analysis of evolving scam tactics. Our team
focused on integrating this AI-driven evaluation with the user interface and community
forum to provide a comprehensive, user-friendly experience. Through this process, we
learned that prompt engineering is crucial for creating adaptable AI systems, and effective
communication.
