Yuchen Jin· @Yuchenj_UW · X·· 2 小时前AI 评分28
AI 导读
我怀疑许多“氛围编程”产品存在的安全漏洞远超其构建者的想象。 AI 让写代码变便宜了，但并没有让写安全的代码变便宜。 安全性应当成为 RL 的一等目标。这可能是某个 neolab 专攻的好方向。
正文 · 原文
I suspect many vibe-coded products have far more security vulnerabilities than their builders realize.
AI made writing code cheap, but it hasn't made writing secure code cheap.
Security should be a first-class RL objective. Could be a great area for a neolab to specialize in.
来源：Yuchen Jin · x.com