Lawsuit says OpenAI contractors read ChatGPT user conversations
Decrypt reports two California ChatGPT users sued OpenAI, claiming outside contractors read their conversations without disclosure.
Why it matters
If the case proceeds, the case puts a consent question in front of a federal court: whether a general model-improvement setting counts as disclosure that user chats are read by human reviewers. A court ruling on default human review would touch how AI companies are allowed to use private conversations for training data. Its significance for crypto markets is limited; the story concerns OpenAI and consumer AI, not digital assets.
Key facts
Two California ChatGPT users filed a proposed class action against OpenAI in the U.S. District Court for the Northern District of California.
Reported by Decrypt
The complaint alleges OpenAI let outside contractors read real user conversations without proper disclosure.
Reported by Decrypt
OpenAI was served on September 2 and has until October 13 to respond in court.
Reported by Decrypt
The complaint concerns an internal OpenAI program called Project Lily, which 404 Media first exposed on September 14.
Reported by Decrypt
Contractors hired through third-party staffing firms for roles such as AI data reviewer and chatbot evaluator read real ChatGPT prompts and full conversations.
Reported by Decrypt
The plaintiffs want the 'Improve the model for everyone' setting switched off by default and consent required before any human reads a chat.
Reported by Decrypt
What happened
Decrypt reports that two California ChatGPT users filed a proposed class action against OpenAI this month in the U.S. District Court for the Northern District of California, alleging the company routed real user conversations to outside contractors without proper disclosure. OpenAI was served on September 2 and has until October 13 to respond. The complaint centers on an internal program Decrypt calls Project Lily, which 404 Media first exposed on September 14. Per the complaint, contractors hired through third-party staffing firms as AI data reviewers and chatbot evaluators read user prompts and full conversations, summarized what the user was trying to do, and scored and critiqued four different model responses on a one-to-seven scale. The plaintiffs want the 'Improve the model for everyone' setting switched off by default, with consent required before any human reads a chat. No response from OpenAI is described in the evidence, and only Decrypt covers the story.
How the story developed
- Update
404 Media exposes internal OpenAI program
404 Media first exposed the internal OpenAI initiative the complaint calls Project Lily, in which staffing-firm contractors read user chats and score model responses.
- First report
First report by Decrypt
Humans Are Reading Your ChatGPT Chats, New Lawsuit Claims
- Update
OpenAI served with the complaint
OpenAI was served on September 2, according to the complaint described by Decrypt.
Affected entities
Original sources
| Publisher | Report | Role | Published |
|---|---|---|---|
| Decrypt Crypto media | Humans Are Reading Your ChatGPT Chats, New Lawsuit Claims | Original |