OpenAI Contractors Reportedly Review Real ChatGPT Conversations

OpenAI is facing renewed scrutiny over privacy after reports that contractors are being hired to review real conversations between users and ChatGPT. The reported programme, known as “Project Lily,” involves human reviewers examining conversations to help assess and improve AI systems, according to reports published on September 14, 2026.

The development has raised concerns because ChatGPT conversations can sometimes contain highly personal or sensitive information. Reports say hundreds of contractors may be involved in reviewing prompts, creating questions about how much access human reviewers have and what safeguards are in place to protect users’ information.

Human review is not entirely new in the development of AI systems. Companies have historically used people to evaluate model responses, identify problems and improve safety. However, the growing sophistication and popularity of conversational AI means that the privacy implications are becoming increasingly important, particularly when users treat AI assistants as private spaces for discussing personal matters.

The controversy also comes as AI companies face broader pressure to be more transparent about how user data is handled. OpenAI and other major AI developers are increasingly introducing privacy-focused systems and policies, while regulators and technology researchers continue to debate what level of human access should be permitted when AI companies process private conversations.

For users, the reports serve as a reminder to think carefully before entering highly sensitive information into any online AI service. While AI companies employ privacy and security measures, no digital platform should automatically be treated as a completely confidential human conversation. The debate surrounding Project Lily is likely to increase pressure on OpenAI to clearly explain how conversations are reviewed, what information contractors can access and what protections are available to users.