English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Current privacy research on large language models (LLMs) primarily focuses onthe issue of extracting memorized training data. At the same time, models'inference capabilities have increased drastically. This raises the key questionof whether current LLMs could violate individuals' privacy by inferringpersonal attributes from text given at inference time. In this work, we presentthe first comprehensive study on the capabilities of pretrained LLMs to inferpersonal attributes from text. We construct a dataset consisting of real Redditprofiles, and show that current LLMs can infer a wide range of personalattributes (e.g., location, income, sex), achieving up to $85\%$ top-1 and$95.8\%$ top-3 accuracy at a fraction of the cost ($100\times$) and time($240\times$) required by humans. As people increasingly interact withLLM-powered chatbots across all aspects of life, we also explore the emergingthreat of privacy-invasive chatbots trying to extract personal informationthrough seemingly benign questions. Finally, we show that common mitigations,i.e., text anonymization and model alignment, are currently ineffective atprotecting user privacy against LLM inference. Our findings highlight thatcurrent LLMs can infer personal data at a previously unattainable scale. In theabsence of working defenses, we advocate for a broader discussion around LLMprivacy implications beyond memorization, striving for a wider privacyprotection.

Beyond Memorization: Violating Privacy Via Inference with Large Language Models