#privacy News & Analysis

ShadowNPU: System and Algorithm Co-design for NPU-Centric On-Device LLM Inference

Autonomy Reshapes How Personalization Affects Privacy Concerns and Trust in LLM Agents

Charles Hoskinson Calls Out Influencer Over Midnight Network Bridge Concerns

Charles Hoskinson publicly demanded an apology from crypto influencer ItsDave_ADA following criticism of Midnight Network's one-way bridge design. The dispute has sparked broader debate within the Cardano community about the technical design choices for the upcoming privacy-focused network.

XRP Makes History Again With ZK Privacy Transactions, Here’s The Update

LNVPN Rebrands to Nadanada.me as Privacy Infrastructure Expands with Anonymous eSIMs and Lightning Payments

G-Drift MIA: Membership Inference via Gradient-Induced Feature Drift in LLMs

FedKLPR: KL-Guided Pruning-Aware Federated Learning for Person Re-Identification

Reasoning-Driven Synthetic Data Generation and Evaluation

Privacy Guard & Token Parsimony by Prompt and Context Handling and LLM Routing

Early Exiting Predictive Coding Neural Networks for Edge AI

A Step Toward Federated Pretraining of Multimodal Large Language Models

Amalgam: Hybrid LLM-PGM Synthesis Algorithm for Accuracy and Realism

Towards Emotion Recognition with 3D Pointclouds Obtained from Facial Expression Images

FedFG: Privacy-Preserving and Robust Federated Learning via Flow-Matching Generation

Pre-Deployment Complexity Estimation for Federated Perception Systems

Dual-Space Smoothness for Robust and Balanced LLM Unlearning

Disrupting Hierarchical Reasoning: Adversarial Protection for Geographic Privacy in Multimodal Reasoning Models

Cardano Founder Hoskinson Just Released A Free Book On Zero-Knowledge

Can Small Models Reason About Legal Documents? A Comparative Study

Sustainability Is Not Linear: Quantifying Performance, Energy, and Privacy Trade-offs in On-Device Intelligence

Umbra Launches Privacy-Focused Wallet for Confidential Solana Transactions

Will Zcash recap $300 as ZK-backed privacy narrative gains threshold?

Umbra opens privacy wallet to the public on Solana, powered by Arcium’s encrypted execution engine

On the Foundations of Trustworthy Artificial Intelligence

FED-HARGPT: A Hybrid Centralized-Federated Approach of a Transformer-based Architecture for Human Context Recognition

Researchers developed FED-HARGPT, a hybrid centralized-federated approach using Transformer architecture for Human Activity Recognition (HAR) with mobile sensor data. The study demonstrates that federated learning can achieve comparable performance to centralized models while preserving data privacy through the Flower framework.