375 articles tagged with #privacy. AI-curated summaries with sentiment analysis and key takeaways from 50+ sources.
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.
Researchers propose Talaria, a new confidential inference framework that protects client data privacy when using cloud-hosted Large Language Models. The system partitions LLM operations between client-controlled environments and cloud GPUs, reducing token reconstruction attacks from 97.5% to 1.34% accuracy while maintaining model performance.
Researchers found that machine unlearning in large language models, which aims to remove specific training data influence, is less effective in interactive settings than previously thought. Knowledge that appears forgotten in static tests can often be recovered through multi-turn conversations and self-correction interactions.
Researchers have identified significant privacy risks in Large Language Model-based Task-Oriented Dialogue Systems, demonstrating that these AI systems can memorize and leak sensitive training data including phone numbers and complete dialogue exchanges. The study proposes new attack methods that can extract thousands of training dialogue states with over 70% precision in best-case scenarios.
Researchers propose Attention Smoothing Unlearning (ASU), a new framework that helps Large Language Models forget sensitive or copyrighted content without losing overall performance. The method uses self-distillation and attention smoothing to erase specific knowledge while maintaining coherent responses, outperforming existing unlearning techniques.
NEAR Protocol's token surged 17% following the launch of 'Confidential Intents', a privacy-focused execution layer designed to prevent MEV and front-running attacks. The token has now gained 40% over the week, outperforming other privacy-focused cryptocurrencies despite modest on-chain revenue generation.
Apple is reportedly asking Google to set up dedicated servers for a new Gemini-powered version of Siri that meets Apple's privacy requirements. This builds on their January partnership announcement where Google's Gemini AI models would help power Apple's upgraded Siri, indicating Apple's increasing reliance on Google's AI infrastructure.
The crypto market enters the week of March 2 with five key catalysts to monitor: escalating US-Iran tensions under Trump, Starknet's privacy-focused Bitcoin wrapper launch, Polygon's March 4 gas upgrade for agentic payments, Avalanche's new incentive program, and Friday's US jobs report.
Cake Wallet has integrated Bitcoin Lightning Network support using Breez SDK and a privacy-focused Spark implementation. The integration enables instant Bitcoin payments while maintaining full self-custody and enhanced privacy by limiting on-chain visibility.
Researchers propose an efficient unsupervised federated learning framework for anomaly detection in heterogeneous IoT networks that preserves privacy while leveraging shared features from multiple datasets. The approach uses explainable AI techniques like SHAP for transparency and demonstrates superior performance compared to conventional federated learning methods on real-world IoT datasets.
Ethereum co-founder Vitalik Buterin announced that native smart accounts (account abstraction) will arrive within a year as part of the Hegota upgrade, after being in development for over a decade. The upgrade will enable users to pay transaction fees in tokens other than ETH and eliminate intermediaries for privacy protocols like Railgun.
Crypto mixer usage has rebounded to near pre-ban levels following the 2022 Tornado Cash sanctions, reaching 32,000 transactions in 2025. Research shows the crackdown primarily deterred legitimate privacy-seeking users while criminals adapted by migrating to alternative platforms and methods.
Small AI models are emerging as a potential solution for private AI applications while fully homomorphic encryption remains years away from frontier-scale deployment. The threshold for what constitutes 'good enough' privacy-preserving AI has been lowered, making smaller models more viable for practical use cases.
A developer is calling for Ethereum to implement Poseidon Hash precompiles to improve zkEVM privacy capabilities and catch up with Solana's development progress. The claim suggests Ethereum is missing this key cryptographic tool that could enhance zero-knowledge Ethereum Virtual Machine functionality.
Paradex is launching as a privacy-first perpetual futures DEX on Layer-2 with institutional-grade performance and zero fees for retail traders. The platform plans to launch its native DIME token in late Q1 2026 with a 25% community airdrop and aims to expand into a comprehensive onchain financial hub.
Starknet announced the launch of a privacy-focused Bitcoin-based asset on its Ethereum layer-2 network, featuring Zcash-like privacy capabilities. This development aims to enhance transaction privacy for users on the scaling solution.
RAILGUN is integrating with Gnosis Guild's Mech standard to enable non-atomic private DeFi transactions. This integration will allow NFT-based mechanisms to enhance privacy capabilities in decentralized finance protocols.
Starknet is preparing to launch strkBTC, a privacy-focused Bitcoin solution for DeFi transactions on its network. STRK token has declined nearly 70% over 90 days and is currently trading between key support at $0.04 and resistance at $0.045.
Cambridge University research shows Railgun has become the most widely used cryptocurrency mixing protocol, while Tornado Cash has experienced modest recovery following the lifting of sanctions. Users are migrating to new platforms as the coin mixer ecosystem recovers from previous regulatory challenges.
Starknet has launched strkBTC, a Bitcoin wrapper that adds privacy features to DeFi transactions without compromising performance. The solution aims to enable private Bitcoin transactions while maintaining composability within the DeFi ecosystem.
Ethereum co-founder Vitalik Buterin sold 17,000 ETH in one month, reducing his holdings from 241,000 ETH to 224,000 ETH according to Arkham data. This selling activity follows his previous announcement of earmarking $45 million for privacy-related initiatives.
Privacy has evolved from a niche concept to essential blockchain infrastructure as public blockchains expose financial data permanently. The next generation of blockchain networks are competing primarily on privacy features rather than transaction speed or fees.
OpenAI is testing advertisements in ChatGPT to support free access to the platform. The company emphasizes that ads will be clearly labeled, maintain answer independence, include strong privacy protections, and provide user control options.
Rui-Siang Lin, founder of Incognito Market, was sentenced to 30 years in prison for operating a cryptocurrency-based dark web marketplace. The platform facilitated $105 million in illegal drug sales, highlighting ongoing regulatory scrutiny of crypto's use in illicit activities.
Microsoft Research has developed OptiMind, a small language model that converts natural language business operation challenges into mathematical formulations for optimization software. The model aims to reduce formulation time and errors while enabling fast, privacy-preserving local deployment.
