Inspiration
The inspiration behind AISecureChat stemmed from the growing need for secure AI Chat platforms in today's digital age. With the increasing prevalence of data breaches and privacy concerns, we wanted to create an AI chatbot solution where user personal identity information data is protected. The inspiration for the project also stems from the US president Executive order cited below
"We need to prioritize the development of privacy-preserving techniques to enable LLMs to be trained on large datasets without the risk of leaking personal details that those datasets might contain." - Presidential Executive Order (Oct 2023)
What it does
AISecureChat is an advanced chatbot powered by artificial intelligence and fortified with state-of-the-art security features by Pangea. It allows users to engage in conversations securely via the Telegram messaging platform. The chatbot leverages AI algorithms to provide intelligent responses while prioritizing user privacy and data protection using Pangea security services
Target Users
This will be used by Developers and Tech Enthusiasts who frequently engage in discussions or collaborations involving sensitive information, such as API keys, passwords, or personal data. Also used by Teams and Organizations who prioritize data security and privacy in their communications.
Use Case of the Bot
This bot stops passing the sensitive PII data being passed on to the large language models (llms) . This prevents data leakage and helps mitigate this risk by automatically identifying and removing such sensitive content, thereby reducing the likelihood of data breaches or unauthorized access.
How I built it
To develop AISecureChat, we embarked on a journey that began with familiarizing ourselves with the Pangea SDK and Telegram Bot creation. Our first step involved exploring the Pangea SDK, a powerful toolset for building secure and scalable applications. We delved into the documentation & exploring pangea-sdk python client for purpose of this project. In this hackathon project, we focussed on the security of protecting user generated prompt that contains personal identification informations. Coming to the second step, we have created telegram bot using bot father telegram app and secured the token. We have used telebot python library to access the chat message and generate the AI prompt completion for the given prompt. Finally integrated it with the llm provider client - fireworks-ai
Challenges I ran into
I have ran the challenge of hosting the telegram bot using vercel.
Accomplishments that I'm proud of
We are very proud of creating the telegram bot using the pangea and fireworks-client python sdk despite i am running under weather.
What I learned
Now, we are confident and learned that whenever we need the secured services we can rely on pangea security services
What's next for AISecureChat
Looking ahead, we envision further enhancing AISecureChat with additional features to improve it as AI Agent for multiple use cases where protecting all user information such as personal data, api keys, financial data is paramount to us.


Log in or sign up for Devpost to join the conversation.