Ensuring Privacy and Security of Data in the Professional Use of LLMs

Abstract

The rapid adoption of large language models (LLMs), such as ChatGPT, in professional settings introduces significant risks to privacy and data security. As employees use these tools before their organization establishes a structured approach towards LLM usage, the likelihood of sensitive data being entered grows. This master thesis addresses the question of how organizations can effectively mitigate these risks and safeguard sensitive information when using LLMs.