← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8788.a · 生成模型:GPT-6

安全左移贯穿整个交付

理解等级:中级

在方案形成时即进行威胁建模、失败模式分析、权限设计和依赖检查,之后将可自动验证的控制接入流水线并持续更新。安全不是上线前一次签字。个人数据需要按用途实施访问控制、适当去标识化与保护,不能把简单匿名化措辞当作不存在再识别风险的保证。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Shifting left on security means incorporating security measures early and throughout the ML development process, rather than considering them as an afterthought or a final checkpoint before deployment. This proactive approach helps teams preempt and defend against vulnerabilities and security issues before they become deeply embedded in the system, making them more costly and difficult to address later.

    参考来源

    - 位置:《Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)》第09章《MLOps与持续交付》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)/_epub-src/OEBPS/ch09.html) - 结论依据:摘录中关于安全左移贯穿整个交付的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:Shifting left on security means incorporating security measures early and throughout the ML development process, rather than considering them as an afterthought or a final checkpoint before deployment. This proactive approach helps teams preempt and defend against vulnerabilities and security issues before they become deeply embedded in the system, making them more costly and difficult to address later.