Training data consent, surveillance concerns and the technical fixes for private AI.
369 articles
The article explores the transition from cloud-based services like ChatGPT to Ollama, a local AI solution. It highlights the advantages of privacy, cost-effectiveness, and offline accessibility for users.
ZDNet AI · 116d ago
A long-standing legal battle regarding users' rights to modify smart TV software is heading to trial. The outcome could impact how users control tracking and advertising on their devices.
Ars Technica · 122d ago
A new report reveals that 38 major data collectors, including AI and defense firms, utilize manipulative design patterns in their opt-out forms. These 'dark patterns' are intended to confuse users and hinder their ability to protect personal data.
Wired · 122d ago
Apple is planning a significant overhaul of Siri, with a strong emphasis on user privacy. Reports suggest that the new version may include features for automatically deleting chat history.
TechCrunch AI · 125d ago
Apple is reportedly introducing a more advanced, chatbot-like Siri in iOS 27 that offers customizable auto-deletion for chat histories. Users can choose to retain conversations for 30 days, one year, or permanently, positioning privacy as a key differentiator.
The Verge · 125d ago
Apple is developing a new ChatGPT-like Siri application that emphasizes user privacy through auto-deleting chat history. Additionally, the company is planning a Genmoji upgrade for the upcoming iOS 27 release.
Bloomberg Technology · 125d ago
The article explores the alarming ease with which facial recognition tools can link professional headshots to non-consensual deepfake content. It highlights the intersection of privacy, identity theft, and the misuse of AI-generated imagery.
MIT Technology Review · 128d ago
Meta has launched 'Incognito Chat', a new feature designed to allow users to interact with Meta AI privately. This mode ensures that chat sessions remain invisible to others, enhancing user privacy during AI interactions.
Meta Newsroom · 129d ago
NHS England allows external contractors including Palantir unrestricted access to identifiable patient data in its flagship data platform's modules. This involves the NDIT, a data safe haven before pseudonymization.
cnBeta (Full RSS) · 131d ago
ChatGPT learns about the world while prioritizing user privacy by reducing personal data in training datasets. It provides users with control over whether their conversations are used to improve AI models.
OpenAI News · 136d ago
ETH Zurich researchers analyzed 62,090 real ChatGPT conversations from 668 users. They trained an AI to predict personality traits from chat history.
Digital Trends · 136d ago
Google Chrome has silently downloaded a 4GB local AI model for Gemini Nano without user consent. This causes unexpected disk space loss and increased I/O activity.
cnBeta (Full RSS) · 136d ago
Reports indicate Apple is developing Siri AI Recaps for newer Apple Watch models, summarizing key events and conversations from a user’s day. The feature reportedly processes audio on-device without storing recordings or transcripts, but its always-on listening model raises significant privacy and consent concerns.
Computerworld · 10d ago
AI coding assistants now use convention files, repository maps, memory banks, and other tools to stay grounded in codebases. The article examines how a similar portable personal context layer could extend AI assistance to areas such as health and finances.
The Next Web (TNW) · 11d ago
IFA 2026 showcases a new generation of screenless wearables that replace displays with sensors, microphones, and cameras. Products such as Luna Band, Memoket Gem, Plaud One, and ODYSS N1 collect health data, conversations, or visual context and increasingly connect these capabilities to AI subscriptions.
极客公园 · 11d ago
摩爾線程 AI Cube 與 Lenovo AI 主機 P7 將本地大模型推理、儲存與家庭設備協作帶入消費市場。文章認為，家庭確實需要端側 AI，但分散在手機、攝影機、NAS 與穿戴裝置中的需求，未必足以支撐一台高價獨立算力設備。
虎嗅 · 12d ago
China’s smart glasses market grew 85.5% year over year in unit sales during the first half of 2026, supported by more than 70 new products, subsidies, and lower prices. However, online return rates reportedly reach 30% and sometimes 70%, highlighting unresolved problems in product usefulness, privacy, safety, battery life, and comfort.
虎嗅 · 12d ago
Millions of children are returning to school with AI chatbots that can complete homework and answer questions they may hesitate to ask adults. The article examines the privacy, trust, and safety implications of children treating chatbots as confidential advisers.
GeekWire · 15d ago
A RIZAP employee mistakenly uploaded customer personal and sensitive health information to an external generative AI service used privately. The exposed data reportedly included names, medical conditions, and insurance card numbers, prompting an apology from the company.
ITmedia AI+ (日本) · 15d ago
Ollie is positioning its family-focused AI assistant around privacy as a competitive advantage. Although it wants access to users’ everyday-life details, the company says it will not use that data to train AI models or share it with others.
TechCrunch AI · 16d ago
Flock’s AI Search lets police officers search across multiple cameras using a written description of a person. WIRED reconstructed the tool from code delivered to an officer’s browser, revealing how the system supports camera-based person searches.
Wired AI · 16d ago
Dating apps are increasingly adopting facial recognition and biometric scanning to deter scammers and issue “verified human” badges. The approach may improve trust, but it is also raising concerns about surveillance, privacy, and biometric-data handling.
Wired · 16d ago
Governor Kathy Hochul discussed AI regulation, data centers, surveillance, teen platform restrictions, and the trade-offs of online age verification. She also addressed New York’s one-year moratorium on data center construction and what would make future projects acceptable.
The Verge · 19d ago
Plaud has announced the $250 Plaud One wireless earbuds, which include 4G connectivity, conversation recording, and an AI assistant that can act on spoken content. Pre-orders are available in eight countries, including Germany and France, where recording private speech without consent can carry criminal penalties.
The Next Web (TNW) · 23d ago
The article argues that Apple’s expansion of Siri AI is creating a new conflict involving regulation, litigation, and data protection. It frames these issues as a strategic challenge for Apple as it develops AI-powered features.
ZDNet AI · 23d ago
Early testers praise Instinct’s capabilities, but its extensive access and ability to act on users’ behalf raise privacy and security concerns. The assistant’s broad terms of service may create uncomfortable trade-offs for users and organizations.
TechCrunch AI · 26d ago
California and 28 other states accuse Meta of designing Facebook and Instagram to be addictive, violating children’s privacy laws, and concealing its practices. Prosecutors describe the alleged strategy as “hook, hold, harvest, hide.”
The Guardian Technology · 28d ago
ChatGPT's new Mac plugin can analyze iMessages to reveal patterns in how users communicate with others. The feature offers a potentially useful way to understand conversations, while also raising important privacy and consent considerations.
ZDNet AI · 29d ago
The ChatGPT macOS app reportedly requests Full Disk Access and permission to access Messages. The permissions could enable deeper system integration, but they also raise significant privacy and data-exposure concerns.
Computerworld · 29d ago
The article reports that ChatGPT can access text messages through app-level integrations. It raises concerns about how much personal data AI assistants may obtain from apps and calls on Apple to strengthen its safeguards.
Ifanr (爱范儿) · 29d ago