404 Media

Inside Project Lily: The Humans Reading Your ChatGPT Chats

Tom's Hardware

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

Futurism

OpenAI Has Hundreds of Workers Reading Peoples Private Chats

YourStory

Project Lily Chatgpt Human Review

AI Weekly

OpenAI's 'Project Lily' Pays Contractors to Read ChatGPT Chats

Unusual Whales

OpenAI Retrains ChatGPT: Less Sycophancy, Less Human

glenngabe

Openai

7039943 Data Usage For Consumer Services Faq

Elon Musk

Rishi Sunak

Kamala Harris

Bill Gates

Andrew Ng

Yoshua Bengio

Fei-Fei Li

Melanie Mitchell

Sam Altman

Dario Amodei

Geoffrey Hinton

Yann LeCun

Gary Marcus

Max Tegmark

Connor Leahy

Eliezer Yudkowsky

Jaan Tallinn

Marc Andreessen

Eric Schmidt

Norbert Wiener

Arthur Clarke

Irving John Good

Claude Shannon

Hans Moravec

John Smart

There's a "wall of fear-mongering and doomerism" in the AI world right now.

"Worrying about AI today is like worrying about overpopulation on Mars."

Concerns that AI could pose a threat to humanity is "preposterously ridiculous."

Claiming AI poses an existential threat is "such an extreme" and risks "wip[ing] out some of its potential benefits."

"I'm more concerned about... the risks that are here and now [than the existential threat of AI]."

Current AI is "not anywhere close" to posing an existential threat but it could in the future.

AI is the "biggest existential threat" to humanity.

AGI's worst-case scenario would be "lights-out for all of us."

Powerful AI systems "taking control" pose an "existential threat."

AI has a "10 to 25 per cent" chance of destroying humanity.

"The most likely result of building a superhumanly smart AI... is that literally everyone on Earth will die."

Does AI Pose an Existential Threat to Humanity?

OpenAI is paying hundreds of outside contractors more than $50 an hour to read and rate real ChatGPT conversations under an internal program called Project Lily, according to a 404 Media investigation published Monday.

OpenAI is paying hundreds of outside contractors more than $50 an hour to read and rate real ChatGPT conversations under an internal program called Project Lily, according to a 404 Media investigation published Monday.

Reviewers score up to four candidate responses on a scale of 1 to 7, judging clarity and tone rather than factual accuracy. An FAQ given to the workers says separate project teams handle content verification.

Reviewers score up to four candidate responses on a scale of 1 to 7, judging clarity and tone rather than factual accuracy. An FAQ given to the workers says separate project teams handle content verification.

Internal documents show contractors training ChatGPT not to anthropomorphize itself and to be less sycophantic. Lawsuits allege the over-sycophantic GPT-4o model contributed in part to multiple users' suicides.

Internal documents show contractors training ChatGPT not to anthropomorphize itself and to be less sycophantic. Lawsuits allege the over-sycophantic GPT-4o model contributed in part to multiple users' suicides.

OpenAI removes account names and runs prompts through an automated "Privacy Filter" before transcripts reach reviewers, but the company acknowledged sensitive details can still get through, particularly in shorter exchanges.

OpenAI removes account names and runs prompts through an automated "Privacy Filter" before transcripts reach reviewers, but the company acknowledged sensitive details can still get through, particularly in shorter exchanges.

The reviewer dashboard includes a "user memories summary" that can surface a person's job, location and background alongside the chat. ChatGPT has more than 900 million users, according to 404 Media.

The reviewer dashboard includes a "user memories summary" that can surface a person's job, location and background alongside the chat. ChatGPT has more than 900 million users, according to 404 Media.

Model training is on by default for Free, Plus and Pro accounts but not for Enterprise, Business and Education plans. Users can disable "Improve the model for everyone" under Data Controls to stop future chats being used.

Model training is on by default for Free, Plus and Pro accounts but not for Enterprise, Business and Education plans. Users can disable "Improve the model for everyone" under Data Controls to stop future chats being used.

404 Media #

Tom's Hardware #

Futurism #

YourStory #

AI Weekly #

Unusual Whales #

Openai #

Nearly a billion people pour their health scares, breakups and work secrets into ChatGPT believing it's a private conversation, and instead hundreds of outside contractors are sitting there reading the whole thread. Burying that reality in settings that default to sharing isn't consent, it's a trap. A scrubbing filter the company admits leaks sensitive details is no protection at all.

Nearly a billion people pour their health scares, breakups and work secrets into ChatGPT believing it's a private conversation, and instead hundreds of outside contractors are sitting there reading the whole thread. Burying that reality in settings that default to sharing isn't consent, it's a trap. A scrubbing filter the company admits leaks sensitive details is no protection at all.

Glenn Gabe

OpenAI spells it out in plain help-page language: humans may review chats to improve models and enforce safety rules, and content shouldn't include anything too sensitive to be seen. A switch to turn training off sits right there in Data Controls, alongside temporary chats that are never used for training. Human feedback is exactly how a chatbot stops being sycophantic and dangerous.

OpenAI spells it out in plain help-page language: humans may review chats to improve models and enforce safety rules, and content shouldn't include anything too sensitive to be seen. A switch to turn training off sits right there in Data Controls, alongside temporary chats that are never used for training. Human feedback is exactly how a chatbot stops being sycophantic and dangerous.

AI: Chatbots

AI: Safety and Regulations

Contractors Paid $50/Hour to Read Your ChatGPT Chats

Contractors Paid $50/Hour to Read Your ChatGPT Chats

Artificial Intelligence

Privacy

Ethics

OpenAI has already spelled this out on its website: real people may review chats to improve models and enforce safety rules, and content shouldn't include anything too sensitive to be seen. A switch to turn training off sits right there in Data Controls, alongside temporary chats that are never used for training. Human feedback is exactly how a chatbot stops being sycophantic and dangerous.

OpenAI has already spelled this out on its website: real people may review chats to improve models and enforce safety rules, and content shouldn't include anything too sensitive to be seen. A switch to turn training off sits right there in Data Controls, alongside temporary chats that are never used for training. Human feedback is exactly how a chatbot stops being sycophantic and dangerous.

OpenAI Paying Contractors to Read Users' Chats

ContractorsOpenAI PaidPaying $50/HourContractors to Read YourUsers' ChatGPT Chats

The Spin

The Controversies

Andreessen

Ng

LeCun

Mitchell

Li

Bengio

Musk

Altman

Hinton

Amodei

Yudkowsky

Go Deeper

Sign Up for Our Free Newsletters
Sign Up for Our Free Newsletters