Long Context, Less Focus: A Scaling Gap in LLMs Revealed through Privacy and Personalization

TL;DR

This paper introduces PAPerBench, a large-scale benchmark that systematically evaluates how increasing context length in LLMs affects personalization quality and privacy risks, revealing a consistent degradation in both as context grows.

Contribution

The paper presents a new benchmark and comprehensive analysis demonstrating a scaling gap in LLMs, where longer contexts lead to reduced personalization and increased privacy leakage, supported by theoretical insights.

Findings

Performance degrades in personalization with longer contexts

Privacy risks increase as context length grows

Attention dilution explains the scaling gap in LLMs

Abstract

Large language models (LLMs) are increasingly deployed in privacy-critical and personalization-oriented scenarios, yet the role of context length in shaping privacy leakage and personalization effectiveness remains largely unexplored. We introduce a large-scale benchmark, PAPerBench, to systematically study how increasing context length influences both personalization quality and privacy protection in LLMs. The benchmark comprises approximately 29,000 instances with context lengths ranging from 1K to 256K tokens, yielding a total of 377K evaluation questions. It jointly evaluates personalization performance and privacy risks across diverse scenarios, enabling controlled analysis of long-context model behavior. Extensive evaluations across state-of-the-art LLMs reveal consistent performance degradation in both personalization and privacy as context length increases. We further provide a theoretical analysis of attention dilution under context scaling, explaining this behavior as an inherent limitation of soft attention in fixed-capacity Transformers. The empirical and theoretical findings together suggest a general scaling gap in current models -- long context, less focus. We release the benchmark to support reproducible evaluation and future research on scalable privacy and personalization. Code and data are available at https://github.com/SafeRL-Lab/PAPerBench