GoKawiilOpenAI's 'Project Lily' has contractors manually reviewing ChatGPT chat logs
A 404 Media investigation reveals OpenAI runs an internal effort called Project Lily, employing hundreds of contractors paid roughly $50 an hour to read anonymized ChatGPT conversations and rate the quality of the bot's answers. Reviewers check for issues like overly robotic phrasing, sycophancy, and improper anthropomorphizing, following guidelines described as inconsistent and frequently changing.
GoKawiil's interpretation of the reporting above, not reported fact.
Many users treat ChatGPT like a confidant or therapist, so learning that real people—not just algorithms—read these exchanges raises fresh privacy concerns even with anonymization in place. It also shows how much manual human labor still underpins the supposedly automated improvement of AI chatbots, complicating claims about how private user data truly is.
- OpenAI's 'Project Lily' uses contractors to manually grade ChatGPT chat transcripts.
- Reviewers earn about $50/hour but describe the work as repetitive and guided by shifting, contradictory rules.
- Users are largely unaware their private conversations may be read by human reviewers, raising privacy concerns.
Source: tomshardware.com — Bruno Ferreira, 2026-09-15
Published there as: “ChatGPT transcripts are reportedly read by humans to improve responses, including those with personal information — 'Project Lilly' has seen OpenAI hire hundreds of contractors to manually review logs”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.