ai-privacy - 技术专题深度解读

nicedreamzapp / claude-code-local

Run Claude Code 100% on-device with local AI on Apple Silicon. MLX-native Anthropic-API server. 6 fighters incl. Muse-Glimmer 30B (now multimodal — reads images, abliterated), Gemma 4 31B, Qwen 3.5 122B (65 tok/s), DeepSeek V4 Flash (1M ctx). Private, offline, airgap-ready. Built for NDA / legal / healthcare workflows.

macos llama airgap gemma mlx on-device-ai ai-privacy voice-ai apple-silicon ambient-computing browser-agent anthropic local-llm local-ai qwen offline-ai private-ai mlx-lm claude-code abliterated

Updated Oct 8, 2026
Python

kekePower / cognito-ai-search

Powerful search page powered by LLMs and SearXNG

open-source privacy ai mit mit-license ai-search ai-privacy searxng ollama cognito-ai-search search-privacy

Updated May 11, 2026
TypeScript

GaaraZhu / gate

A deterministic privacy boundary between your data and AI.

rust mcp data-governance ai-security ai-privacy privacy-engineering ai-governance llm-security agentic-ai pii-protection

Updated Aug 20, 2026
Rust

ZhengyuZhao / AI-Security-and-Privacy-Events

A curated list of academic events on AI Security & Privacy

adversarial-machine-learning adversarial-examples ai-security ai-privacy data-poisoning

Updated Aug 22, 2024

JGalego / awesome-safety-critical-ai

When the stakes are high, intelligence is only half the equation - reliability is the other ⚠️

machine-learning awesome reliability artificial-intelligence awesome-list compliance reproducibility ai-safety safety-critical awesome-lists determinism explainability responsible-ai ai-privacy trustworthy-ai genai

Updated Oct 1, 2026
JavaScript

AI-secure / aug-pe

[ICML 2024 Spotlight] Differentially Private Synthetic Data via Foundation Model APIs 2: Text

language-model differential-privacy ai-privacy large-language-models prompt-engineering

Updated Jan 11, 2025
Python

NoWayJA / word-to-markdown

Convert Word docs to Markdown privately - 100% offline, no uploads. Perfect for processing sensitive documents with Ollama, LM Studio, GPT4All & other local AI tools. Just double-click the standalone/word-to-markdown.html file to use.

markdown privacy offline docx client-side no-upload document-converter ai-privacy gpt4all local-llm ollama lm-studio word-to-markdown

Updated Feb 4, 2026
TypeScript

flare-collection / flare-redact

Zero-dependency TypeScript secret and PII redaction for any SDK, logs, HTTP, LLM prompts, and scoped MCP tool boundaries.

Updated Aug 9, 2026
TypeScript

CorvusCodex / OfflineAI

Sponsor

OfflineAI is an artificial intelligence that operates offline and uses machine learning to perform various tasks based on the code provided. It is built using two powerful AI models by Mistral AI.

ai offline ai-privacy ai-model computer-ai local-ai offlineai aioffline offline-artificial-intelligence privacyai offline-ai offline-ai-download offline-ai-model no-internet-ai pc-ai artifical-inteligence ai-download download-ai

Updated Jul 27, 2025
Python

AI-Security-Research-Group / LLM-Attacks

Comprehensive taxonomy of AI security vulnerabilities, LLM adversarial attacks, prompt injection techniques, and machine learning security research. Covers 71+ attack vectors including model poisoning, agentic AI exploits, and privacy breaches.

machine-learning ai-security ai-privacy open-source-security adversarial-ml prompt-injection ai-red-teaming model-poisoning llm-attacks ai-vulnerabilities security-taxonomy

Updated Sep 19, 2025

didvc / opencode-bwrap

Run opencode inside a bubblewrap sandbox — confine AI file access to your current working directory

Updated Aug 26, 2026
Shell

pooyaphoenix / RA3G-Agent

Local RAG system with a built-in governance agent that filters sensitive or restricted information with separated agent logging systems to keep privacy and security

agent multi-agent security-tools ai-security rag pii-detection ai-privacy llm ollama streamlit-ai local-rag agentic-ai rag-data rag-security

Updated Aug 1, 2026
Python

fullstackcrew-alpha / privacy-mask

Automatically redacts sensitive data in screenshots before sending to AI agents

Updated Mar 24, 2026
Python

HKU-TASR / Geminio

[ICCV 2025] Geminio is a VLM-powered gradient inversion attack in federated learning (FL). It allows the adversary (the FL server) to describe the data of value and reconstruct the victim client's private data matching the description.

vlm federated-learning ai-privacy gradient-inversion

Updated Aug 7, 2025
Python

tamimalmahmud / LLM-Unlearning

The LLM Unlearning repository is an open-source project dedicated to the concept of unlearning in Large Language Models (LLMs). It aims to address concerns about data privacy and ethical AI by exploring and implementing unlearning techniques that allow models to forget unwanted or sensitive data. This ensures that AI models comply with privacy.