Skip to content
Synced · WeChat

MemPrivacy Shows a Privacy Layer for AI Memory

越懂你,越危险?MemPrivacy揭示AI记忆下一站

MemTensor and HONOR open-sourced MemPrivacy for edge-cloud agent memory protection using local reversible pseudonymization; MemPrivacy-4B-RL reached 85.97% composite F1 on MemPrivacy-Bench, 50.47 percentage points above OpenAI privacy-filter, while the benchmark covers 200 users and more than 155,000 privacy items.

Why it matters: MemTensor and Honor open-sourced MemPrivacy, which swaps sensitive details in agent memory for reversible aliases locally, so the model only sees fake names and the real ones are restored when needed.

Read the original ↗Export Markdown