OpenAI uses contractors to review private ChatGPT conversations

OpenAI is utilizing hundreds of contractors to review real ChatGPT conversations as part of an internal program codenamed ‘Project Lily’. The initiative aims to improve AI responses by having human reviewers evaluate the quality and accuracy of chatbot replies.

While OpenAI states that a privacy filter is used to strip personal information, reports indicate that reviewers can still see user memory summaries, which may contain details about a user's past interactions or general location. Although usernames are not visible, the context provided can sometimes reveal sensitive information such as medical, legal, or financial details.

This practice has drawn regulatory scrutiny. Italy’s regulator previously fined OpenAI EUR 15M for processing data without an adequate legal basis. The controversy highlights ongoing debates regarding the duty of AI companies to inform users when their private conversations are being accessed by human personnel.