--- title: Hundreds of contractors are reportedly reading real... description: OpenAI pays hundreds of contractors, under the internal codename Project Lily, to read real ChatGPT conversations and rate the chatbot's replies, according to... canonical: https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Hundreds of contractors are reportedly reading real ChatGPT conversations | daily.dev og:description: OpenAI pays hundreds of contractors, under the internal codename Project Lily, to read real ChatGPT conversations and rate the chatbot's replies, according to... og:url: https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt og:image: https://api.daily.dev/og/posts/yEkyk39YT.png og:image:alt: Hundreds of contractors are reportedly reading real ChatGPT conversations og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Hundreds of contractors are reportedly reading real ChatGPT conversations **[The Next Web](https://daily.dev/sources/tnw)** · 3 min read · 0 upvotes · 0 comments ## Summary OpenAI pays hundreds of contractors, under the internal codename Project Lily, to read real ChatGPT conversations and rate the chatbot's replies, according to a 404 Media report. Reviewers cannot see usernames but can see a memory summary that may reveal what a user has used the chatbot for and where they might live. OpenAI's Privacy Filter model is meant to strip personal information first, but the company's own documentation admits it can miss identifiers or under-redact when context is limited. OpenAI would not tell 404 Media where it discloses to users that humans may read their chats, pointing only to a help page after publication. Anthropic and Google Gemini also use human review, with different opt-in defaults and disclosure language. The EU Court of Justice ruled in EDPS v SRB that a controller's duty to inform users applies at the moment of data collection, regardless of whether the recipient can identify the person, which is relevant since Italy's regulator already fined OpenAI EUR 15 million, partly for processing without an adequate legal basis. The 'improve the model for everyone' setting that feeds human review is on by default for free, Plus, and Pro accounts and off for Enterprise, Business, and Edu; disabling it only affects new conversations, not past ones. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Does OpenAI let humans read ChatGPT conversations for training and review purposes? Yes, OpenAI pays contractors under the internal codename Project Lily to read real ChatGPT conversations and rate the chatbot's replies. Reviewers don't see usernames but can see a memory summary that may reveal prior chatbot usage and rough location. A Privacy Filter model strips personal information first, but OpenAI's own documentation says it can miss uncommon identifiers or under-redact when context is limited. _Track how AI vendors handle chat data and disclosure obligations by following privacy and GDPR coverage on daily.dev._ ### Is the 'improve the model for everyone' setting on by default in ChatGPT? Yes, it's enabled by default for free, Plus, and Pro accounts, and disabled by default for Enterprise, Business, and Edu accounts. Turning it off only stops future conversations from being used; conversations already sent before switching it off remain part of the training and review pipeline. _Developers weighing which AI tools to trust with user data can compare privacy defaults on daily.dev._ ### What did the EU Court of Justice rule about a company's duty to inform users before sharing their data with third parties? The Court of Justice held in EDPS v SRB that a controller's duty to inform people about data transfers applies at the moment of collection, judged from the controller's own perspective rather than the recipient's ability to identify anyone. This means whether a contractor could identify who wrote a prompt is irrelevant; the disclosure obligation exists regardless, and Italy's regulator has already fined OpenAI EUR 15 million partly on related grounds. _Engineers building on GDPR-regulated AI platforms can follow compliance rulings like this via daily.dev._ ## Similar posts on daily.dev - [OpenAI wants you to use AI — but not to train its AI](https://daily.dev/posts/openai-wants-you-to-use-ai-but-not-to-train-its-ai-9ozzebxa6) · InfoWorld · 0 upvotes · 0 comments - [We still don’t know how people are really using AI](https://daily.dev/posts/we-still-don-t-know-how-people-are-really-using-ai-tl6darxnq) · MIT Technology Review · 0 upvotes · 0 comments - [I didn't realize ChatGPT was serving ads until I dug into how it actually works](https://daily.dev/posts/i-didn-t-realize-chatgpt-was-serving-ads-until-i-dug-into-how-it-actually-works-iywrshaes) · XDA Developers · 2 upvotes · 0 comments - [Study: AI chatbots provide less-accurate information to vulnerable users](https://daily.dev/posts/study-ai-chatbots-provide-less-accurate-information-to-vulnerable-users-8nuiign7m) · MIT News · 1 upvotes · 0 comments --- Tags: [#openai](https://daily.dev/tags/openai), [#chatgpt](https://daily.dev/tags/chatgpt), [#data-privacy](https://daily.dev/tags/data-privacy), [#ai-safety](https://daily.dev/tags/ai-safety), [#gdpr](https://daily.dev/tags/gdpr) [View this post on daily.dev](https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Hundreds of contractors are reportedly reading real ChatGPT conversations","url":"https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt"},"datePublished":"2026-09-14T17:16:44.761Z","dateModified":"2026-09-14T17:34:26.876Z","description":"OpenAI pays hundreds of contractors, under the internal codename Project Lily, to read real ChatGPT conversations and rate the chatbot's replies, according to...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/3995a9f2ed463be16314ab91d5f4b24e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/3995a9f2ed463be16314ab91d5f4b24e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"openai,chatgpt,data-privacy,ai-safety,gdpr","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"Hundreds of contractors are reportedly reading real ChatGPT conversations"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/hundreds-of-contractors-are-reportedly-reading-real-chatgpt-conversations-yekyk39yt#faq","mainEntity":[{"@type":"Question","name":"Does OpenAI let humans read ChatGPT conversations for training and review purposes?","acceptedAnswer":{"@type":"Answer","text":"Yes, OpenAI pays contractors under the internal codename Project Lily to read real ChatGPT conversations and rate the chatbot's replies. Reviewers don't see usernames but can see a memory summary that may reveal prior chatbot usage and rough location. A Privacy Filter model strips personal information first, but OpenAI's own documentation says it can miss uncommon identifiers or under-redact when context is limited. Track how AI vendors handle chat data and disclosure obligations by following privacy and GDPR coverage on daily.dev."}},{"@type":"Question","name":"Is the 'improve the model for everyone' setting on by default in ChatGPT?","acceptedAnswer":{"@type":"Answer","text":"Yes, it's enabled by default for free, Plus, and Pro accounts, and disabled by default for Enterprise, Business, and Edu accounts. Turning it off only stops future conversations from being used; conversations already sent before switching it off remain part of the training and review pipeline. Developers weighing which AI tools to trust with user data can compare privacy defaults on daily.dev."}},{"@type":"Question","name":"What did the EU Court of Justice rule about a company's duty to inform users before sharing their data with third parties?","acceptedAnswer":{"@type":"Answer","text":"The Court of Justice held in EDPS v SRB that a controller's duty to inform people about data transfers applies at the moment of collection, judged from the controller's own perspective rather than the recipient's ability to identify anyone. This means whether a contractor could identify who wrote a prompt is irrelevant; the disclosure obligation exists regardless, and Italy's regulator has already fined OpenAI EUR 15 million partly on related grounds. Engineers building on GDPR-regulated AI platforms can follow compliance rulings like this via daily.dev."}}]} ```