New Locket Framework Enhances LLM Privacy and Access Control

Researchers have developed a new framework called Locket that enhances security and privacy for large language models (LLMs). Locket uses lightweight adapters, each associated with a specific access policy, and a gating module that employs learned keyed entry tokens to route requests. This system allows for fine-grained control over access to private data, enabling features like PII masking or differential privacy, without significantly compromising model utility. Experiments show Locket maintains performance comparable to fine-tuning on raw data when authorized, while effectively reducing PII leakage when unauthorized. AI

IMPACT Enhances LLM security and privacy, potentially enabling wider adoption in sensitive domains.

RANK_REASON Academic paper detailing a new technical framework for LLM security. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Enron
• European Court of Human Rights
• Gemma 2-2B
• Llama~3.2
• LLMs
• Locket
• LoRA
• Meta*
• Mohamed Elmahallawy
• PII
• Qwen3
• Yelp

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →