AI 每日简报

技术创新 OpenAI 发布 Codex Security 研究预览版，采用 AI 驱动的约束推理和验证技术，替代传统静态应用安全测试（SAST），能够以更少的误报发现复杂漏洞。日本电商巨头 Rakuten 已采用 Codex 编码助手，将平均修复时间（MTTR）降低 50%，自动化 CI/CD 审查，数周内完成全栈构建。

争议焦点 MIT 科技评论深度分析 OpenAI 与五角大楼协议后的军事应用争议。文章指出，OpenAI 技术可能在伊朗冲突中用于目标分析和打击优先级排序，AI 模型可分析文本、图像、视频等多种情报输入，向人类分析师提供打击建议。这是生成式 AI 首次在实战中被测试用于行动建议。

Technological Innovation OpenAI launched a research preview of Codex Security, an AI-powered alternative to traditional SAST that detects complex vulnerabilities with fewer false positives. Japan's Rakuten adopted Codex, slashing its MTTR by 50% and automating CI/CD reviews for faster full-stack development.

Controversy Focus MIT Technology Review analyzed controversy over OpenAI's Pentagon agreement: its AI was reportedly tested in the Iran conflict to analyze intelligence (text, images, video) and advise on target strikes—marking generative AI's first real-combat operational testing.