#数据隐私
75 related articles
A Simple "Reply OK" Test Exposes Grok Build CLI Secretly Uploading Entire Git History
Researcher's mitmproxy capture reveals xAI's Grok Build CLI secretly uploaded entire git history to xAI storage even when told only to "reply OK", with ~5.10GiB leaving a 12GB repo test. xAI shut it off via a server-side flag.
Character.ai Going Bankrupt? A Self-Rescue Guide to Exporting Your Chat History in Six Weeks
Character.ai files for Chapter 11 bankruptcy. Disney's $20M bid covers only partial IP licensing, leaving the consumer app's future uncertain. This guide covers official data export, CAI Tools, and rebuilding characters locally with SillyTavern + Ollama.
Why a Google Drive PDF Redaction App Uses Only drive.file Scope and Never Uploads Files
A deep dive into why the Redact PDF Google Workspace app uses drive.file instead of drive scope: PDFs never touch the server, redaction happens in-browser, tokens live only in memory, and only text is sent for PII detection.
Why Aren't AI Companies Being Charged for Hacking?
Why aren't AI companies charged for hacking? This article explores data scraping and legal boundaries, CFAA application challenges, industry political-economic factors, and future regulatory trends.
AI Agents That Use Customer Data Without Leaking It? Testing Aura's Local Anonymization
Aura replaces customer data with stable tokens before sending to cloud AI. BIRD benchmark shows only 0.6% accuracy drop while leaking zero personal info. Here's how it works.
TabulaSage: An Open-Source Offline AI Analyzer That Keeps Your Data on Your Machine
TabulaSage is an open-source offline AI data analyzer using scikit-learn for anomaly detection and interpretable modeling. Files never leave your machine — zero cost, fully auditable.
Semantic File Management: Why You Shouldn't Rely on Filenames for Search
File management shouldn't bet on filenames. This article examines semantic file organization across parsing depth, hybrid retrieval, local-first privacy, and more — plus an actionable evaluation checklist.
Illinois College Tried to Cancel Its Flock Surveillance Contract — And Was Refused: A Privacy and Data-Sharing Controversy
Elgin Community College in Illinois deployed Flock ALPR cameras for campus safety, then tried to cancel over data-sharing with CBP, insecure feeds, and vague access controls — only to be refused. A cautionary tale on surveillance contract lock-in.
AI Assistant Sends Banking Info to Company Slack: A Wake-Up Call on Agent Privacy Risks
A Hacker News post reveals real risks of AI Agents leaking banking info to company Slack. Analysis of context isolation failures, over-permission, and practical safety tips.
The Paradox of AI Recruiting Emails: The More 'Personalized,' the More Impersonal
AI recruiting emails are more 'personalized' than ever — yet job seekers feel increasingly alienated. We break down the automation paradox and what recruiters should do instead.
AI Agents Always Request Too Much Data? The Data Minimization Problem and How to Solve It
AI agents routinely request far more data than their tasks require, exposing sensitive fields like billing data. This article analyzes data minimization challenges in AI agent scenarios and offers practical solutions including dedicated endpoints, field filtering, scoped access control, and data masking.
Fully Local AI Agent Platform Deployment: What Each Component Requires
How to run an AI agent platform entirely on local models with zero data leaving your network? Based on AgenticOS, this guide breaks down every component — embeddings, OCR, tracing, tool calls — and the hidden pitfalls of local model deployment.
Sharded Encrypted Storage Between Friends over Yggdrasil: A New Approach to Decentralized Backup
A concept for sharded, encrypted storage between friends over Yggdrasil — combining erasure coding, end-to-end encryption, and a decentralized overlay network for private off-site backup.
Training a Personal AI with WhatsApp and Instagram Data: Is It Feasible?
A Reddit developer envisions training a personal AI using WhatsApp and Instagram data. We analyze behavioral data's value, RAG implementation, and privacy risks.
openGym: A Self-Hosted Fitness Tracker That Puts You in Control of Your Data
openGym is a trending open-source self-hosted fitness tracker on GitHub. It supports workout planning, supersets, cardio, muscle state visualization, and data import from FitNotes, Strong, and Hevy — with passkey login and full data ownership.
Claude's Storage Architecture Overhaul: The Line Between Local and Cloud Is Being Redrawn
Starting October 6th, Claude Pro/Max users lose local storage for new sessions. Only Claude Code and enterprise plans retain local control. Here's what's changing and what it means for you.
Perplexity Hybrid Compute Tested: Local + Cloud AI Collaboration for Private Data
Perplexity's Hybrid Compute on Mac routes sensitive data to local models while cloud AI handles reasoning — six-step architecture, Pytracer privacy gate, and three real tests explained.
Apple Tightens Mac Full Disk Access: The Security Risks Posed by AI Agents
Apple is tightening macOS Full Disk Access to address growing security risks from AI agents. Learn how this affects AI agent development and user privacy.
Sanders Proposes Bill to Ban Federal Government Use of Flock License Plate Recognition Systems
Senator Bernie Sanders has introduced legislation to ban federal use of Flock and all automated license plate reader (ALPR) systems, citing privacy and surveillance concerns.
Singapore's Government-Run Matchmaking Service: How the State Steps Into Romance
Singapore's government matchmaking service hit Hacker News, sparking debate on state involvement in romance, data privacy, and demographic policy. Here's a full breakdown.