Martin Gubri

A tool to check personal information safety in large language models.

APRICOT enhances trust in language models by measuring answer confidence accurately.

A look into membership inference attacks and their relevance in data privacy.