Claude 程式碼外洩附惡意軟體

#code-leak#malware#supply-chain-attackclaudeclaudefbicisco

💡Claude 外洩+惡意軟體：開發者勿下載未驗證 AI 程式碼。

⚡ 30 秒速覽

有什麼變化

駭客張貼感染惡意軟體的 Claude 原始碼外洩

為什麼重要

暴露 Claude 架構潛在漏洞供惡意利用。強調 AI 企業供應鏈風險，呼籲小心處理程式碼。可能促使 Anthropic 加強安全措施。

下一步行動

使用 VirusTotal 掃描所有非官方 AI 原始碼後再分析。

誰應關注：Developers & AI Engineers

本篇為 AI 生成分析，非原文內容。

•The malicious payload is primarily distributed via compromised GitHub repositories and unofficial developer forums, masquerading as a 'Claude Code' CLI tool to exploit developers' trust in Anthropic's ecosystem.
•Security researchers have identified the malware as a sophisticated infostealer designed to harvest environment variables, API keys, and local SSH credentials, specifically targeting developers working with LLM-integrated workflows.
•Anthropic has issued an official advisory clarifying that 'Claude Code' is not a publicly released open-source project, urging users to only interact with tools via their official API documentation and verified distribution channels.

📊 競品分析▸ Show

Feature	Anthropic (Claude)	OpenAI (o1/GPT-4)	Google (Gemini)
Primary Interface	Web/API/Claude Code (Official)	Web/API/OpenAI CLI	Web/API/Google AI Studio
Pricing	Usage-based (API)	Usage-based (API)	Usage-based (API)
Developer Focus	High (Prompt Caching/Tool Use)	High (Reasoning Models)	High (Multimodal/Agentic)