GitHub 发布开放 AI 代码审查基准 ReviewBench
ReviewBench is a benchmark for code review agents built on representative GitHub pull requests, multi-source ground truth, and calibrated evaluation.
ReviewBench 是一个面向代码审查智能体的基准，基于有代表性的 GitHub 拉取请求、多来源真实标注和校准评估构建。
原文来源：GitHub Blog（github.blog）。本页摘要与要点整理自公开来源，版权归原作者所有，完整内容请阅读原文。
Google froze its open source bug bounty program due to a ‘significant rise’ in AI submissionsAI 生成的漏洞报告激增，谷歌暂停开源漏洞赏金计划
TechCrunch · 2026-10-05
LibreOffice says ‘no AI’ is now a software featureLibreOffice 宣布默认不加入 AI 功能
TechCrunch · 2026-10-06
Building Git infrastructure for agent-scale developmentGitHub 分享面向“智能体规模”开发的 Git 基础设施建设思路
GitHub Blog · 2026-10-07
Apple says it’s tightening macOS ‘Full Disk Access’ controls due to new risks from AI agents苹果收紧 macOS“完全磁盘访问”权限，应对 AI 智能体带来的新风险
TechCrunch · 2026-10-03
Highlights from Git 2.56Git 2.56 发布，GitHub 解读重点新特性
GitHub Blog · 2026-09-29
Figma announces new features to define how agents work on its platformFigma 发布智能体协作新功能，代码图层进入公开测试
TechCrunch · 2026-10-15