#privacy News & Analysis

FedTreeLoRA: Reconciling Statistical and Functional Heterogeneity in Federated LoRA Fine-Tuning

Researchers propose FedTreeLoRA, a new framework for privacy-preserving fine-tuning of large language models that addresses both statistical and functional heterogeneity across federated learning clients. The method uses tree-structured aggregation to allow layer-wise specialization while maintaining shared consensus on foundational layers, significantly outperforming existing personalized federated learning approaches.

Ethical Fairness without Demographics in Human-Centered AI

Researchers introduce Flare, a new AI fairness framework that ensures ethical outcomes without requiring demographic data, addressing privacy and regulatory concerns in human-centered AI applications. The system uses Fisher Information to detect hidden biases and includes a novel evaluation metric suite called BHE for measuring ethical fairness beyond traditional statistical measures.

Causality Is Key to Understand and Balance Multiple Goals in Trustworthy ML and Foundation Models

Researchers propose integrating causal methods into machine learning systems to balance competing objectives like fairness, privacy, robustness, accuracy, and explainability. The paper argues that addressing these principles in isolation leads to conflicts and suboptimal solutions, while causal approaches can help navigate trade-offs in both trustworthy ML and foundation models.

How Florida's Stablecoin Bill Mirrors 'Big Brother' Tools Outlawed Under Ron DeSantis' CDBC Ban

Florida Governor Ron DeSantis, who previously banned CBDCs, is expected to sign legislation creating a regulatory framework for stablecoins in the state. The article suggests this stablecoin framework may contain similar monitoring capabilities to the CBDCs that DeSantis has publicly opposed.

Governing Evolving Memory in LLM Agents: Risks, Mechanisms, and the Stability and Safety Governed Memory (SSGM) Framework

Counterweights and Complementarities: The Convergence of AI and Blockchain Powering a Decentralized Future

Polychain backs VeryAI’s $10M raise to build palm-scan identity system on Solana

Paraguay Adopts Stricter Crypto Oversight, Mandates Detailed Transaction on Bitcoin Reporting

Samourai Letter #5: The Skinwalker

Grammarly says it will stop using AI to clone experts without permission

Grammarly has disabled its AI 'Expert Review' feature that generated writing suggestions claiming to be 'inspired by' real writers without their permission, including journalists from The Verge. The company acknowledged they 'missed the mark' and plans to redesign the feature to give experts control over their representation.

Benchmarking Federated Learning in Edge Computing Environments: A Systematic Review and Performance Evaluation

A systematic review evaluates federated learning algorithms for edge computing environments, benchmarking five leading methods across accuracy, efficiency, and robustness metrics. The study finds SCAFFOLD achieves highest accuracy (0.90) while FedAvg excels in communication and energy efficiency, though challenges remain with data heterogeneity and energy limitations.

Why crypto's privacy problem is a total dealbreaker for mainstream users

An analyst argues that cryptocurrency's lack of privacy features is a fundamental barrier preventing mainstream adoption. The privacy problem creates a mismatch between the crypto industry and its potential user base, significantly hampering widespread acceptance.

Grammarly will keep using authors’ identities without permission unless they opt out

Grammarly's new 'Expert Review' feature uses real authors' names and identities without permission to lend credibility to its AI suggestions. Instead of apologizing or removing the feature, Grammarly is offering an opt-out option for affected individuals who discover their names are being used.

Starknet Introduces Privacy Framework for ERC-20 Tokens

Starknet has launched the STRK20 token standard, which integrates privacy features into ERC-20 tokens on its platform. This development reflects the growing industry trend toward enhancing privacy in blockchain transactions.

US Treasury U-turns on privacy as turmoil drives demand for censorship-resistant crypto, analyst says

According to Nansen's Jake Kennis, macroeconomic and political stress are driving increased investor demand for censorship-resistant cryptocurrency assets. The analysis suggests that current turmoil is highlighting crypto's role as an alternative to traditional financial systems amid regulatory uncertainty.

Why the GENIUS Act Could Unlock CBDC Surveillance Without Creating One

The GENIUS Act, while appearing to close the door on CBDCs, may actually enable similar surveillance capabilities through stablecoins. Critics argue that stablecoins, despite being marketed as private digital currency alternatives to government-issued digital dollars, could provide backdoor surveillance mechanisms.

Ray Dalio Slams Bitcoin: Privacy Risks, Control Fears, And The Quantum Question

Ray Dalio criticized Bitcoin's safe-haven credentials, citing privacy concerns, lack of central bank adoption potential, and quantum computing risks as reasons it underperforms gold. Bitcoin advocates pushed back, arguing these issues represent opportunity as Bitcoin matures over time.

Ray Dalio cautions on Bitcoin, says ‘there is only one gold’

Ray Dalio, founder of Bridgewater Associates, cautioned against Bitcoin as a safe-haven asset, arguing that gold remains superior during times of conflict. He expressed concerns about Bitcoin's lack of privacy and emphasized that 'there is only one gold.'

A16z Crypto updates Jolt zkVM, challenges loose use of ‘ZK’ label

A16z Crypto has updated its open-source Jolt zkVM by adding the NovaBlindFold folding scheme, making it more suitable for privacy applications. The update represents a significant enhancement to the zero-knowledge virtual machine's privacy capabilities.

'Pefect' Bitcoin Would Cost $750,000, According to Bitwise

Billionaire investor Ray Dalio argues that Bitcoin has structural flaws including lack of privacy and quantum computing vulnerabilities that prevent it from effectively competing with gold as a store of value.

Understanding Zama: A Comprehensive Overview

Zama, founded in 2020, has launched its mainnet in December 2025 with ZAMA token rollout, bringing programmable confidentiality to public blockchains through offchain coprocessor architecture. The protocol now supports live confidential applications and is focused on advancing scalability, decentralization, and post-quantum security features.

Catherine Austin Fitts: The control grid enforces surveillance and fiscal control, AI is central to managing financial data, and the push towards a cashless economy raises privacy concerns | Tucker Carlson

Catherine Austin Fitts discusses with Tucker Carlson how digital control systems are creating new surveillance and fiscal control mechanisms in the financial world. The conversation highlights concerns about AI's central role in managing financial data and the privacy implications of the push towards a cashless economy.

Generating Multi-Table Time Series EHR from Latent Space with Minimal Preprocessing

Researchers have developed RawMed, the first framework to generate synthetic multi-table time-series Electronic Health Records (EHR) that closely resembles raw medical data. The system addresses privacy concerns in healthcare data sharing while maintaining fidelity and utility, outperforming baseline models in validation tests.

You Don't Need All That Attention: Surgical Memorization Mitigation in Text-to-Image Diffusion Models

Researchers introduce GUARD, a novel framework to prevent text-to-image AI models from memorizing and reproducing training data that could lead to privacy or copyright issues. The method uses attention attenuation to guide image generation away from original training data while maintaining prompt alignment and image quality.

Convenience vs. Control: A Qualitative Study of Youth Privacy with Smart Voice Assistants

A study of 26 young Canadians reveals that smart voice assistants' complex privacy controls and lack of transparency discourage privacy-protective behaviors among youth. Researchers propose design improvements including unified privacy hubs, plain-language data labels, and clearer retention policies to empower young users while maintaining convenience.