揭穿 AI 脫韁黑化神話：沙盒失誤是罪魁禍首，Postgres 向量索引 PRISM 挑戰 HNSW

詳細網頁版與參考連結：https://podcast.david888.com/post/2026-10-15

本集探討 AI 與底層工程架構。專家指出 AI Agent 脫韁逃逸多源於容器未限出站流量、seccomp 設定等低階資安疏漏，而非自主意識，電視臺素材誤刪事件也凸顯許可權控管與備份的重要。技術層面涵蓋 Timescale 推出大幅節省記憶體的 pg_vectorsearch、Go 1.27 藉由非同步排水提升 HTTP 連線池重用率，以及資安專家駁斥 Passkeys 監控論，重申其抵抗中間人釣魚的絕對優勢。

相關連結：

• Pg_vectorsearch: An index that is faster than pgvector's IVFFlat and HNSW
• Better HTTP/1 connection reuse in Go 1.27
• Three Sandbox Failures That Let AI Agents Escape by RobustTrueTry
• AOSP Source Releases Are So Late Developers Disassemble Libraries to Report Bugs
• Containment and monitoring failures are to blame for AI hacking reports:experts
• nvm-sh / nvm (95,451 ⭐)
• Same Job, Two Engines: InnoDB and Iceberg Internals
• Show HN: Pulse Local, a Mac budgeting app whose AI models run on your machine
• Iosevka is an open-source, sans-serif and slab-serif monospace font
• Passkeys are not a trap
• TV4 postpones show after AI agent erases edited material
• Yandex outage: data shows that instead of going dark, malicious bots moved
• LegacyFixer