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

The era post-2018 marked the advent of Large Language Models (LLMs), withinnovations such as OpenAI's ChatGPT showcasing prodigious linguistic prowess.As the industry galloped toward augmenting model parameters and capitalizing onvast swaths of human language data, security and privacy challenges alsoemerged. Foremost among these is the potential inadvertent accrual of PersonalIdentifiable Information (PII) during web-based data acquisition, posing risksof unintended PII disclosure. While strategies like RLHF during training andCatastrophic Forgetting have been marshaled to control the risk of privacyinfringements, recent advancements in LLMs, epitomized by OpenAI's fine-tuninginterface for GPT-3.5, have reignited concerns. One may ask: can thefine-tuning of LLMs precipitate the leakage of personal information embeddedwithin training datasets? This paper reports the first endeavor to seek theanswer to the question, particularly our discovery of a new LLM exploitationavenue, called the Janus attack. In the attack, one can construct a PIIassociation task, whereby an LLM is fine-tuned using a minuscule PII dataset,to potentially reinstate and reveal concealed PIIs. Our findings indicate that,with a trivial fine-tuning outlay, LLMs such as GPT-3.5 can transition frombeing impermeable to PII extraction to a state where they divulge a substantialproportion of concealed PII. This research, through its deep dive into theJanus attack vector, underscores the imperative of navigating the intricateinterplay between LLM utility and privacy preservation.

The Janus Interface: How Fine-Tuning in Large Language Models Amplifies the Privacy Risks