Fact-Forensics - AI Agent for Deepfake Detection

The deepfake threat has evolved from a hypothetical concern into a significant financial and social risk. With the market for deepfake technology projected to reach $7.27 billion by 2031 and fraud losses already exceeding $410 million in the first half of this year, the need for effective solutions is more urgent than ever. Current detection tools are falling short; human accuracy in spotting deepfakes is as low as 0.1%, while even leading AI models have limitations. Fact-Forensics is an intelligent agent designed to operate on the Coral Protocol agents ecosystem to combat this challenge head-on. Built with LangChain, our project is designed to be LLM provider agnostic, functioning with services like OpenAI, Mistral, or Gemini. This modularity allows us to leverage specialised capabilities, such as using Mistral's advanced image understanding to analyse uploaded content and Tavily for web searches to cross-reference information online. Our innovation empowers citizens directly, building a more trustworthy online environment.