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 impressive performances of large language models (LLMs) and their immensepotential for commercialization have given rise to serious concerns over theintellectual property (IP) of their training data. In particular, the synthetictexts generated by LLMs may infringe the IP of the data being used to train theLLMs. To this end, it is imperative to be able to (a) identify the dataprovider who contributed to the generation of a synthetic text by an LLM(source attribution) and (b) verify whether the text data from a data providerhas been used to train an LLM (data provenance). In this paper, we show thatboth problems can be solved by watermarking, i.e., by enabling an LLM togenerate synthetic texts with embedded watermarks that contain informationabout their source(s). We identify the key properties of such watermarkingframeworks (e.g., source attribution accuracy, robustness against adversaries),and propose a WAtermarking for Source Attribution (WASA) framework thatsatisfies these key properties due to our algorithmic designs. Our WASAframework enables an LLM to learn an accurate mapping from the texts ofdifferent data providers to their corresponding unique watermarks, which setsthe foundation for effective source attribution (and hence data provenance).Extensive empirical evaluations show that our WASA framework achieves effectivesource attribution and data provenance.

WASA: WAtermark-based Source Attribution for Large Language Model-Generated Data