#privacy News & Analysis

An AI Agent Execution Environment to Safeguard User Data

Introducing OpenAI Privacy Filter

Coinbase Study Finds Privacy Tools Like Railgun Are Mathematically Safe From Quantum Attacks

Distributed Human Identity: AI-Enabled Multi-Existence Through Cognitive Replication and Robotic Embodiments

Rethinking Post-Unlearning Behavior of Large Vision-Language Models

Stop Tracking Me! Proactive Defense Against Attribute Inference Attack in LLMs

Toward Zero-Egress Psychiatric AI: On-Device LLM Deployment for Privacy-Preserving Mental Health Decision Support

Privacy-Preserving LLMs Routing

Researchers propose PPRoute, a privacy-preserving framework for LLM routing that uses Secure Multi-Party Computation (MPC) to protect user data while dynamically selecting between model providers. The system achieves 20x speedup over naive MPC implementations through optimized encoder inference, multi-step model training, and an efficient Top-k algorithm, maintaining routing quality without sacrificing privacy.

Synthetic data in cryptocurrencies using generative models

Researchers propose using Conditional Generative Adversarial Networks (CGANs) to generate synthetic cryptocurrency price data, addressing privacy and access concerns in financial research. The approach combines LSTM generators with MLP discriminators to produce statistically consistent synthetic time series that preserve market dynamics, offering a computationally efficient alternative for financial modeling and analysis.

Tempo’s ‘Zones’ sparks debate over privacy in crypto infrastructure

Tempo has launched 'Zones,' a feature enabling private, permissioned stablecoin transactions on its layer-1 blockchain targeting enterprises. The development has triggered criticism from the crypto community, with detractors arguing that the operator-controlled architecture reintroduces centralized trust mechanisms, potentially undermining core decentralization principles.

Blockchain Is South Korea’s New Fiscal Weapon — A Blow To Privacy?

Tempo Unveils 'Zones' for Private Enterprise Stablecoin Transactions

ZK-APEX: Zero-Knowledge Approximate Personalized Unlearning with Executable Proofs

From Anchors to Supervision: Memory-Graph Guided Corpus-Free Unlearning for Large Language Models

You Switched to Claude Over Surveillance Fears. Now It Wants Your Passport

Anthropic has implemented government ID and selfie verification for Claude users, marking the first major AI chatbot to adopt such measures. This move contradicts the company's recent privacy-focused positioning that attracted users fleeing ChatGPT, raising questions about the tension between identity verification and user privacy.

PrivacyReasoner: Can LLM Emulate a Human-like Privacy Mind?

Researchers introduce PrivacyReasoner, an LLM-based agent architecture that reconstructs individual privacy perspectives from online comment history to predict how specific people would perceive data practices. The system outperforms baseline models in predicting privacy concerns across AI, e-commerce, and healthcare domains by contextually activating relevant privacy beliefs.

Why companies like Apple are building AI agents with limits

Apple, Qualcomm, and other tech companies are developing next-generation AI agents intentionally designed with built-in limitations rather than unrestricted capabilities. These agents can perform tasks like app navigation, bookings, and service management, but operate within controlled parameters that prioritize safety and user privacy over maximum autonomy.

AlphaTON raises $43 million to build sovereign AI and privacy infrastructure

Transparent When You Want It, Private When You Need It

Zcash Price Surges Over 30% in 24 Hours as Grayscale Accumulates $46 Million in Shielded ZEC

Solana’s privacy narrative sharpens as SOL Strategies buys Darklake Labs

Crypto Trust Crisis — The “Kim Jong‑Un Test” Is Exposing Secret North Korean Moles

Bitcoin price-drop speculation spurred by familiar price pattern

Selective Forgetting for Large Reasoning Models

Researchers propose a new framework for 'selective forgetting' in Large Reasoning Models (LRMs) that can remove sensitive information from AI training data while preserving general reasoning capabilities. The method uses retrieval-augmented generation to identify and replace problematic reasoning segments with benign placeholders, addressing privacy and copyright concerns in AI systems.

AI agents, privacy and prediction markets define ETHGlobal Cannes 2026 finalists

ETHGlobal Cannes 2026 announced 10 finalists featuring projects focused on AI agents, privacy infrastructure, and on-chain prediction markets. Notable projects include ENShell, DIVE, Corpus, and VEIL VPN, representing some of the most technically advanced submissions in ETHGlobal's history.