📄 中文摘要
随着AI生成代码的涌入，软件开发的瓶颈正从编写转向审查、测试和确保机器生成代码的安全性。Anthropic观察到，Claude Code的代码库中约90%的代码是由Claude Code自身生成的，工程师的角色转变为监督而非手动编写。这种规模打破了传统的审查和责任假设。在整个行业中，84%的开发者正在使用或计划使用AI编码工具，约42%的提交代码是AI生成的。在如此高的生成量下，自动化审查中的漏洞变得系统性。
Anthropic的新Claude代码审查：自动化AI时代的软件质量
出处: Anthropic S New Claude Code Review Automating Ai Age Software Quality
发布: 2026年3月16日
📄 English Summary
Anthropic S New Claude Code Review Automating Ai Age Software Quality
As AI-generated code increasingly fills repositories, the bottleneck in software development is shifting from writing to reviewing, testing, and securing machine-generated outputs. Anthropic has observed that approximately 90% of Claude Code's own codebase is now generated by Claude Code itself, with engineers taking on a supervisory role rather than authoring code by hand. This scale disrupts traditional assumptions about review and accountability. Across the industry, 84% of developers are using or planning to use AI coding tools, and around 42% of committed code is AI-generated. At such high volumes, gaps in automated review become systemic.