Dimitri Staufer, Kirsten Morehouse
This study audits how large language models (LLMs) associate personal data with individuals, revealing the models' ability to accurately generate personal information and raising privacy concerns.
Researchers investigated how large language models, like those used in chatbots, associate personal information with people's names. They developed a tool to audit these associations and found that the models could accurately generate details such as gender and languages spoken for well-known individuals, and even for everyday users to some extent. Most participants in the study wanted more control over what information the models associate with their names, highlighting concerns about privacy and data protection. This raises important questions about how personal data is handled by these models and whether privacy laws should apply to them.
