Phantoms

Scammers are getting creative with their scam tactics. ScamQuest educates you about the latest, modern scams whether it's about Taylor Swift tickets or new CDC vouchers, so we're always aware.

Comment

Inspiration

One of the major challenges in combating financial frauds is the constantly evolving tactics used by fraudsters. For instance, during the rollout of government initiatives like the CDC vouchers, fraudsters have exploited the public's eagerness to claim benefits by setting up fake websites and phishing for personal information. To address this, ScamQuest continuously scrapes the latest information from local forums and websites in Singapore. By scraping such current trends and feeding this data into a Large Language Model (LLM), we generate questions that reflect the most recent scams and tactics. This ensures that users are always aware of the latest threats and can recognize them in real-time, constantly being tested on the most relevant information and exposed to the most recent types of frauds.

What it does

ScamQuest is a gamified way of learning about the most modern scams in Singapore. The questions are updated with the latest scams, so you're always vigilant on scams to look out for.

How we built it

Frontend: The frontend is developed using Next.js and React. Backend: We used Flask to handle the backend operations. Question Generation: We utilized Langchain and OpenAI to generate questions based on the latest scraped data. Web Scraping: Selenium and Beautiful Soup were used for scraping relevant data from local forums and websites, ensuring our content remains current and relevant.

Challenges we ran into

Being able to engineer our prompts in order to output consistent results that we want turned out to be a big challenge for us. We wanted the LLMs to not only come up with convincing scams, but also ones that are commonly exploited amongst Singaporeans. Eventually we got satisfactory results, and we're excited to see the potential of this application.

Accomplishments that we're proud of

In this short period of time, we were able to produce a comprehensive full stack application integrated with an LLM!

What we learned

Prompt engineering is difficult, but powerful. Our team also experimented with new things and learnt a lot about fullstack development

What's next for Phantoms

  • Honing our gamification strategy
  • Further improving the realism of our scam examples.
  • Predictions of possible scams based on modern news (currently only tackles reported scams)

Built With

Share this project:

Updates