Prakhar Ganesh, Yash More, Marco Romanelli, Ferdinando Fioretto, Golnoosh Farnadi · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.29925 · cited by 0 works (OpenAlex) · open-access PDF
Da große Sprachmodelle (LLMs) im Umfang und in der Fähigkeit weiter wachsen, steigen Datenschutzbedenken. Da Benutzer zunehmend mit einem einzigen allgemeinen Zweck-System in vielen unterschiedlichen Kontexten interagieren, gibt es eine natürliche Tendenz, eine ein-Size-fit-all-Lösung für die Privatsphäre zu verfolgen. Leider übersieht dies einen entscheidenden Einblick aus Jahrzehnten der Privatsphäreforschung: Privatsphäre ist inhärent kontextuell. In diesem Papier argumentieren wir, dass die bestehende Arbeit in der sofortigen Privatsphäre in Richtung Versuche, einen einheitlichen Rahmen zu entwerfen, die sehr irreführend ist und ein falsches Gefühl der Sicherheit fördern kann. Darüber hinaus argumenti
Machine translation (Meta M2M100); the English original follows.
As large language models (LLMs) continue to grow in scope and capability, privacy concerns are on the rise. Since users increasingly interact with a single general-purpose system across many distinct contexts, there is a natural tendency to pursue a one-size-fits-all solution to privacy. Unfortunately, this overlooks a crucial insight from decades of privacy research: privacy is inherently contextual. In this paper, we argue that existing work in prompt privacy is drifting towards attempts to design a unified framework, which is highly misleading and can foster a false sense of security. Furthermore, we argue that even future work in this direction needs to wrestle with fundamental tensions across various desirable properties of prompt privacy. Thus, such continued attempts towards a unified framework need to be reconsidered. We recommend three critical paths forward: application-specific deployment of LLMs, context-specific evaluation of privacy frameworks, and better privacy literacy.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Scope (computer science), Internet privacy, Context (archaeology), Information privacy, Software deployment, Privacy by Design
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.29925, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange