斯坦福论文提出本地差分隐私水印,让 AI 水印不再侵犯个体隐私
ChengleiSi · x · 2026-11-06
Anthropic and OpenAI will watermark their responses, but critics argue this violates user privacy and incentivizes paraphrasing away the watermark — a lose for both sides.
A Stanford paper by Elise L. Han and Tatsunori Hashimoto, "Reconciling Privacy with Provenance for Language Model Watermarks," reframes the incentive structure: shift the AI provenance goal from individual tracking to estimating aggregate AI usage. The authors develop locally differentially private watermarks that provably limit the evidence any individual text can contribute to detection, while still enabling accurate estimation of the population prevalence of watermarked texts.
「安全」频道最新
- Polymarket:年内通过 AI 安全法案概率仅 9%,2027 年中前 83% — Polymarket · 2026-11-06
- 男子起诉 OpenAI:称 ChatGPT 助其妻制定投毒杀夫计划 — Polymarket · 2026-11-06
- Krueger 与 Hobbhahn 激辩模型威胁模型:scheming 并非唯一危险 — MariusHobbhahn · 2026-11-06
- Sentinel Bio 谈 AI 生物安全：wet-lab 能力提升实验与 DNA 合成筛查 — The Cognitive Revolution · 2026-11-06
- Sentinel Bio 谈前沿 AI 生物风险：KYC 走向 KYA，DNA 筛查成防线 — The Cognitive Revolution · 2026-11-06
- 访谈预告：中国人为何比美国人更拥抱AI？开源模型背后的算力博弈 — panickssery · 2026-11-06