🚨 Chatbot AI Azure: Detecting Scams, Hoaxes & Gambling Promotions
Chatbot AI Azure: Web app to detect scams, hoaxes, gambling promos with AI-powered analysis in multiple languages.
Chatbot AI: Detecting Scams, Hoaxes & Gambling Promotions
A multilingual AI chatbot that helps identify suspicious messages, hoaxes, and online gambling promotions using natural-language analysis.
1. 📝 Introduction
It all started with a simple question: “What if people could take a suspicious message they got on WhatsApp, Telegram, or SMS and quickly find out if it’s a scam, a hoax, or just a gambling promotion?” As online fraud and misinformation grow more clever, I felt a strong urge to see how AI — especially Azure’s range of services — could be used for real-world safety. That’s how Chatbot AI Azure came to life: a multilingual, AI-driven web app that lets users analyze questionable text messages with just a click.
This initiative blends Azure’s cutting-edge AI services with a user-friendly interface to create a digital “shield” against scams. Whether you’re dealing with a gambling promotion, a phishing attempt, or a viral hoax, this tool can quickly identify threats and explain why a message is dangerous. It’s like having a cybersecurity expert in your pocket, always ready to help you evaluate that too-good-to-be-true offer before you click.
Photo by Hartono Creative Studio on Unsplash
2. 🎯 Background & Objectives
The purpose of this project was to investigate how AI can be used to combat misinformation and fraud in real time. Here are the primary objectives we focused on:
- Detect text messages that seem suspicious, such as scams, hoaxes, or those promoting gambling
- Classify and clarify the risk level using Azure OpenAI’s large language model
- Offer support for multiple languages to enhance global usability (like English, Indonesian, Arabic, etc)
- Create a responsive, minimalist web interface with an easy-to-navigate user experience
- Deploy the project on Microsoft Azure for dependable access and reliability in the real world
3. 🧠 Dataset
Unlike typical machine learning projects that depend on structured datasets, this one takes a different approach. It utilizes zero-shot classification with the GPT model hosted on Azure OpenAI, all guided by a thoughtfully crafted prompt. For tasks like sentiment and language detection, I turned to Azure Cognitive Services — specifically, the Text Analytics API. Just a quick reminder: when generating responses, always stick to the specified language and avoid using any others.
4. 🛠️ Project Stages
The project was divided into smaller, more manageable stages that could be tackled one at a time:
- Ideation: Defined the use case around detecting hoaxes, scams, and gambling promotions
- Frontend Interface: Designed with Tailwind CSS for a clean, responsive UI
- Backend Development: Built using Python Flask to handle requests, interact with Azure APIs, and return structured responses
- Prompt Engineering: Crafted a structured system message for GPT that returns a JSON classification
- Language Detection & Sentiment: Integrated Azure Text Analytics to enrich results with language and emotional context
- Multilingual Support: Implemented fallback strategies for Arabic and RTL languages
- Testing & Logging: Included input validation, timeout handling, and GPT fallback when needed
- Deployment: Deployed to Azure App Service with environment-based configuration using .env and Azure App Settings
5. 🧪 How It Works
Using Chatbot AI Azure is as simple as copy–paste, but under the hood, it combines several intelligent processes.
✨ Here’s how to use it:
- Paste or type any suspicious message — for example, something you received via WhatsApp, SMS, or Telegram.
- Click the “Analyze Now” button.
- In seconds, the app will: — Detect the language of the input — Classify the message into: Potential Scam, Online Gambling Promotion, Hoax or Safe — Provide a clear explanation of the risk — Highlight specific danger indicators — Display the confidence level of the analysis — Analyze the sentiment (Positive, Neutral, or Negative)
How it works
6. 🌐 Live Demo
You can try the live application here: 🔗 Current Version (Gemini API): https://wiradp.github.io/chatbot-ai/ This version integrates Google Gemini API for text classification and is publicly accessible.
🔗 Previous Deployment (Azure API): https://cekfakta-ai-app.azurewebsites.net (Note: this used Microsoft Azure OpenAI during the initial free trial, which has now expired.)
Feel free to test it with: — Scammy giveaway messages — “Situs gacor” gambling promos — Viral hoaxes — Or harmless everyday texts
CekFakta AI Demo
The AI will evaluate and explain the content — helping users make informed decisions instantly
⚠️ Disclaimer
If you encounter the message:
“App is no longer available”
or
“Your subscription has been disabled”
This is because the free Azure subscription used to host the application has expired. As a result, the deployed web app and AI services are no longer active.
You can still run the application by visiting this website https://wiradp.github.io/chatbot-ai/
7. 🔍 Insight
Here are some key insights from this project:
- Prompt quality is everything A well-structured prompt can yield highly consistent zero-shot classification, reducing the need for custom training.
- Multilingual UX adds real value Users frequently encounter scams in their own languages. By integrating language detection and adjusting the output, we can enhance accessibility and increase trust.
- Azure AI Services integrate smoothly with Flask With the right setup, Azure OpenAI and Text Analytics can be integrated with minimal latency and high reliability.
- User feedback loop is critical It was important to manage short messages like “OK” or just single words effectively — adding logic for minimal input and fallback translations really boosted the system’s robustness.
8. ✅ Conclusion
This project highlights how effectively AI can be utilized to solve real-world issues like misinformation and fraud. Here are some important takeaways:
- AI models such as GPT can manage classification tasks without the need for conventional training.
- Deploying on Azure provides the scalability and stability necessary for success.
- Language awareness enriches the experience for a diverse global audience.
- Compact, focused applications can address significant problems with high impact.
9. 🔮 Recommendations & Future Work
✅ Current improvements implemented:
- Multilingual detection and output in the same language as input
- Confidence level and sentiment score included in output
- Responsive UI with animated transitions and error handling
🔭 Future Recommendations:
- WhatsApp / Telegram Bot Integration Extend the app to work directly in popular messaging platforms for real-time scam detection.
- User Feedback System Collect user ratings on classification accuracy to improve prompt engineering and trust.
- Admin Monitoring Dashboard Visualize trends in hoax, scam, and gambling message reports. Help stakeholders see what threats are circulating.
- Image / Link Analysis Capabilities Go beyond text! Analyze suspicious links and images to catch phishing attempts or fake ads.
- Customizable Prompt Management Allow easy updates to the classification rules and explanations without code changes.
- Multi-AI Provider Support Enable switching between AI providers. This project can be modified in the future to use other AI services (paid or free) beyond Azure — such as OpenAI’s official API, Google Vertex AI, Cohere, Anthropic, OpenRouter, or even local LLMs — giving flexibility and cost control.
If you’d like to test the project live: 👉 https://wiradp.github.io/chatbot-ai/
Or explore the codebase on GitHub: 🔗 **https://github.com/wiradp/chatbot-ai-azure**
Thank you for taking the time to read this.
I’m learning to write, and mistakes are inescapable even when I do my hardest. Please feel free to offer feedback and recommendations. Let me know if you spot any difficulties or mistakes 🤝
메타데이터
- post_id
- 8b2901293ccb
- slug
- chatbot-ai-azure-detecting-scams-hoaxes-gambling-promotions-8b2901293ccb
- url
- https://medium.com/@wiradp/chatbot-ai-azure-detecting-scams-hoaxes-gambling-promotions-8b2901293ccb
- canonical_url
- https://medium.com/@wiradp/chatbot-ai-azure-detecting-scams-hoaxes-gambling-promotions-8b2901293ccb
- author_url
- https://medium.com/@wiradp
- status
- ok
- fetched_at
- 2026-07-19 06:24:50
