VerifyFirst

Inspiration

The inspiration for VerifyFirst came from witnessing the widespread circulation of misinformation among students and young adults. In today’s digital world, news travels faster than ever, and it is often difficult to distinguish between reliable and misleading content. Many students rely on social media platforms for news, which frequently presents headlines out of context or sensationalized to attract clicks. This not only affects academic work but also shapes opinions based on false information. The idea behind VerifyFirst was to create a tool that empowers students to quickly and easily verify the credibility of the news they encounter, providing them with a clear assessment and understandable reasoning. The goal was to make fact-checking accessible, intuitive, and reliable while also introducing them to the power of artificial intelligence in a practical, educational setting.

What it does

VerifyFirst is an AI-powered web application that allows students to determine the credibility of any news headline or short article in real time. Users simply paste the text they want to verify into the input box, and the system analyzes it using advanced natural language processing models. The tool then provides a credibility score, a classification such as Likely Real, Uncertain, or Likely Fake, and clear explanations outlining why the news was classified in that way. Each explanation includes factors such as emotional language, the presence or absence of named sources, the structure of the headline, and whether the claim matches known fact-check databases. By combining AI predictions with easily understandable reasoning, VerifyFirst helps students make informed decisions about the news they read or share, reducing the spread of misinformation and promoting critical thinking.

How we built it

We built VerifyFirst using a combination of modern frontend technologies, a robust backend, and artificial intelligence APIs. The frontend was developed using HTML5, Tailwind CSS, and JavaScript to create a responsive, visually appealing interface with a dark theme and neon accents that highlight important information. The backend is powered by Flask, which exposes a REST API endpoint to handle requests from the frontend. The AI layer leverages pretrained HuggingFace NLP models specifically fine-tuned for fake news detection. When a user submits text, the backend sends it to the model to obtain probabilities for fake or real classification, calculates a credibility score, and generates explanations based on both AI predictions and optional fact-check database results. The system was designed to always return a result for any input, ensuring a consistent and reliable experience for the user.

Challenges we ran into

could provide a result for every input, even when the news was very recent or not present in fact-check databases. We had to carefully design fallback logic using AI predictions to prevent the system from failing or returning empty results. Another challenge was balancing accuracy with explainability. While pretrained AI models could provide classifications, generating human-readable explanations that made sense to students required additional rules and logic. Designing a frontend that was both visually appealing and user-friendly on different devices while maintaining the neon dark theme was also a challenge, as it required careful attention to colors, spacing, and interactive elements. Finally, integrating multiple APIs while keeping the system responsive and fast was a technical hurdle that required optimization and efficient coding.

Accomplishments that we're proud of

We are proud of several accomplishments with VerifyFirst. Firstly, we created a system that reliably classifies news as fake, real, or uncertain, while always providing explanations for its decisions. This was achieved in a short time frame, demonstrating both efficiency and technical skill. Secondly, we designed a visually appealing, dark-themed interface that is user-friendly and works across devices, which enhances the user experience and makes the project suitable for a hackathon presentation. We also successfully integrated AI and optional fact-check APIs in a seamless way, ensuring that the tool remains fast and responsive even under multiple requests. Overall, we are proud of building a project that is not only technically sound but also has a meaningful social impact by promoting critical thinking and reducing misinformation among students.

What we learned

Throughout the development of VerifyFirst, we learned a great deal about full-stack development, AI integration, and user-centered design. We deepened our understanding of natural language processing models and how to leverage pretrained models for real-world applications. We gained experience in designing a REST API that communicates efficiently with the frontend, and we learned best practices for creating responsive and visually appealing user interfaces using Tailwind CSS. Additionally, we learned the importance of fallback logic and explainability in AI systems to ensure reliability and trustworthiness. Finally, we gained insights into how to structure a hackathon project to balance technical complexity with clarity, speed, and impact, which will be invaluable for future projects.

What's next for VerifyFirst

The next steps for VerifyFirst involve expanding its capabilities and reach. We plan to develop a browser extension that can verify news directly on social media platforms and messaging apps, allowing users to fact-check information in real time. Multilingual support will also be added, enabling verification of news in different languages such as Urdu and English. We intend to integrate additional fact-check databases and sources to improve confidence and coverage. Future improvements may also include highlighting specific words or phrases in the text that contributed to the AI’s classification, enhancing transparency and learning. Ultimately, we aim to make VerifyFirst a widely accessible tool that helps students and young adults navigate the digital news environment with confidence and critical thinking.

Built With

• css/tailwind
• flask
• html5
• huggingface-transformers
• javascript
• python