Discovery
We study what AI can do.
Independent AI research lab
AI research for discovery, detection & defense.
Our work covers privacy-preserving learning, deepfake detection, and AI systems designed to withstand attacks.
Research system
Three research areas
Detection
We identify AI-generated content.
Defense
We build systems that can withstand attacks.
15
Publications
Peer-reviewed work
8
Projects
Research directions
2
Researchers
Based in Dublin
Selected work
Featured research
A selection of current projects from across the lab.
active
May 2026
UB-SMoE: Universally Balanced Sparse Mixture-of-Experts for Resource-adaptive Federated Fine-tuning
A resource-adaptive federated fine-tuning method that fixes the expert-imbalance and gradient-sparsity discordances of sparse MoE, achieving up to 45% computation reduction and 8.7x better low-resource performance over heterogeneous LoRA-rank methods.
- less computation
- 45%
- better low-resource performance
- 8.7×
federated learningmixture of expertsresource-adaptive
active
Mar 2026
TriDetect: Semi-supervised Generalized AI-generated Image Detection
A semi-supervised detector that learns architectural patterns in fake images and generalizes across image generators.
deepfake detectiongeneralization
active
Nov 2025
A Survey on Proactive Deepfake Defense: Disruption and Watermarking
A survey of disruption and watermarking methods for proactive deepfake defense across visual and audio media.
deepfake defenseproactive detection
Latest updates
Recent activity
New papers, projects, and notes from the lab.
PaperProjectPaperPaperProjectBlogPaperPaperPaper
UB-SMoE: Universally Balanced Sparse Mixture-of-Experts for Resource-adaptive Federated Fine-tuning of Foundation Models
Proceedings of the International Conference on Machine Learning (ICML) · 2026
UB-SMoE: Universally Balanced Sparse Mixture-of-Experts for Resource-adaptive Federated Fine-tuning
A resource-adaptive federated fine-tuning method that fixes the expert-imbalance and gradient-sparsity discordances of sparse MoE, achieving up to 45% computation reduction and 8.7x better low-resource performance over heterogeneous LoRA-rank methods.
AFSS: Artifact-Focused Self-Synthesis for Mitigating Bias in Audio Deepfake Detection
International Joint Conference on Neural Networks (IJCNN) · 2026
Beyond Binary Classification: A Semi-supervised Approach to Generalized AI-generated Image Detection
Proceedings of the AAAI Conference on Artificial Intelligence · 2026
TriDetect: Semi-supervised Generalized AI-generated Image Detection
A semi-supervised detector that learns architectural patterns in fake images and generalizes across image generators.
Introducing GMD-AI: Why we started this group
Why we formed GMD-AI and how discovery, detection, and defense shape our research.
Transferable Multi-Bit Watermarking Across Frozen Diffusion Models via Latent Consistency Bridges
Second Workshop on Technical AI Governance Research (TAIGR 2026) · 2026
How Effective Are Publicly Accessible Deepfake Detection Tools? A Comparative Evaluation of Open-Source and Free-to-Use Platforms
Preprint (arXiv) · 2026
Deepfake Detection Across Image, Video, and Audio: A Comprehensive Survey with Empirical Evaluation of Generalization and Robustness
Springer AI Review · 2026
Showing 9 updates