Particle.news

Download on the App Store

IndiaAI Picks Five Projects to Build Deepfake, Bias and Generative AI Security Tools

A multi-stakeholder panel chose the winners from more than 400 proposals to turn IndiaAI’s safety agenda into practical tools.

Overview

  • IIT Jodhpur and IIT Madras will develop Saakshya, a multi‑agent, RAG‑enhanced framework focused on deepfake detection and governance.
  • IIT Mandi with the Himachal Pradesh Directorate of Forensic Services will build AI Vishleshak to bolster detection of audio‑visual deepfakes and handwritten signature forgeries with adversarial robustness, explainability and domain generalisation.
  • IIT Kharagpur will create a real‑time voice deepfake detection system to counter synthetic speech manipulation.
  • Digital Futures Lab and Karya will evaluate gender bias in agricultural large language models and produce digital public goods for benchmarking and fair data work.
  • Globals ITES and IIIT Dharwad will develop Anvil, a penetration‑testing and evaluation tool for generative AI systems.