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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Machine learning practitioners often fine-tune generative pre-trained modelslike GPT-3 to improve model performance at specific tasks. Previous works,however, suggest that fine-tuned machine learning models memorize and emitsensitive information from the original fine-tuning dataset. Companies such asOpenAI offer fine-tuning services for their models, but no prior work hasconducted a memorization attack on any closed-source models. In this work, wesimulate a privacy attack on GPT-3 using OpenAI's fine-tuning API. Ourobjective is to determine if personally identifiable information (PII) can beextracted from this model. We (1) explore the use of naive prompting methods ona GPT-3 fine-tuned classification model, and (2) we design a practical wordgeneration task called Autocomplete to investigate the extent of PIImemorization in fine-tuned GPT-3 within a real-world context. Our findingsreveal that fine-tuning GPT3 for both tasks led to the model memorizing anddisclosing critical personally identifiable information (PII) obtained from theunderlying fine-tuning dataset. To encourage further research, we have made ourcodes and datasets publicly available on GitHub at:https://github.com/albertsun1/gpt3-pii-attacks

Does fine-tuning GPT-3 with the OpenAI API leak personally-identifiable information?