Tuesday, 26 August 2025
Location: Function Room 3
07:45–08:30 — Registration
08:30–08:40 — Welcome by Chair
08:40–09:10 — Keynote: Combating Deepfakes: From Passive Detection to Active Protection
Prof. Nhien-An Le-Khac, UCD
Abstract:
In recent years, the rapid advancement of generative AI technologies has led to the widespread creation and misuse of deepfakes (DFs) for malicious purposes, including identity impersonation, misinformation dissemination, and unauthorized imitation of artists’ styles. These applications raise serious concerns related to privacy, security, and copyright infringement. In response, a growing number of DF detection methods have been developed to counter these emerging threats.
In this talk, I will first present an overview of passive DF detection techniques, highlighting their performance across various modalities—such as images, audio, and video—and evaluating their accuracy in current literature. I will also examine the underlying threat models for these passive approaches, considering potential adversarial strategies and varying levels of attacker knowledge and capabilities.
Unlike passive methods that detect deepfakes post-creation and distribution, proactive defense strategies aim to prevent the generation of harmful synthetic content at its source. In the second part of the talk, I will briefly explore proactive defense mechanisms, including disruption techniques and digital watermarking. Finally, I will discuss key challenges and outline promising directions for future research in both passive detection and proactive prevention of deepfakes.
09:10–09:30 — Article 1: LLaVA-SNIPPER: Scene-Graph-based Inference with Multimodal LLMs for Explainable Out-of-Context Misinformation
Trong-Thuan Nguyen, Minh-Triet Tran
09:30–09:50 — Article 2: STDD: A Hybrid Spatial-Temporal Model for Deepfake Detection
Thuan Minh Pham, Lam Thu Bui, Hien Van Tran, Trung Duy Pham
09:50–10:10 — Article 3: Enhancing Medical Chatbot Reliability: A Multi-Step Verification Approach to Prevent Hallucinations
Trong-Nghia Tran, Minh-Nhat Nguyen, Minh-Triet Tran, Trong-Le Do
10:10–10:20 — Q&A buffer
10:20–10:40 — Tea Break ☕
10:40–11:00 — Article 4: A Hybrid Model for Generalizable Deepfake Detection via Blending, Semantic, and General Artifacts
Minh-Khoa Le-Phan, Minh-Hoang Le, Minh-Triet Tran, Trong-Le Do
11:00–11:20 — Article 5: E-FreeM2: Efficient Training-Free Multi-Scale and Cross-Modal News Verification via MLLMs
Van-Hoang Phan, Long-Khanh Pham, Dang Vu, Anh-Duy Tran, Minh-Son Dao
11:20–11:50 — Invited Paper: A Practical Framework for Encrypted Medical Prediction using Blockchain-Anchored Key Management System
Rafik Hamza, Minh-Son Dao
11:50–12:30 — Open Discussion / Wrap-up