Meta用实际案例讲隐私控制的难点，一个‘age’字段在不同场景含义不同，搞错了会出大问题。
Meta在博客中分享了构建隐私感知基础设施时，资产分类面临的挑战。隐私控制（如保留、访问、用途限制）需要对数据有准确理解，但相同名称可能在不同上下文有不同含义，例如字段“age”在不同场景代表不同数据。Meta通过案例研究说明了如何设计分类系统来识别这些差异。
Privacy-Aware Infrastructure in the AI-Native Era: An Asset Classification Case Study
Privacy controls — systems that enforce retention, access, allowed-purpose, downstream-sharing, or anonymization policies — require a reliable understanding of data to function. Before such a control can operate effectively, it must know exactly what it is looking at. This can be complex, as demonstrated by a field simply named “age“: In one context, it might [...] Read More... The post Privacy-Aware Infrastructure in the AI-Native Era: An Asset Classification Case Study appeared first on Engineering at Meta .