People Training OpenAI's AI Fired for Using AI to Train the AI
Recorded: Sept. 22, 2026, 2:09 p.m.
| Original | Summarized |
People Training OpenAI’s AI Fired for Using AI to Train the AI
Account Log in Navigation Home About Follow us Twitter
Sign in
About Advertisement Go ad free OpenAI Joseph Cox · Sep 22, 2026 IMAGE: SPUMONI COOPERATIVE FOR 404 MEDIA.
OpenAI has an army of contractors who read real ChatGPT users’ prompts and other data to help improve the chatbot’s responses. The idea is that the contractors provide an, obviously, human touch to OpenAI’s models. Well, not all of the contractors are doing that. 404 Media has found multiple contractors hired to improve OpenAI’s models have been fired for using AI to train the AI. That’s not great for the models themselves, but there is also obviously a great irony in AI training companies working for OpenAI firing people for using AI when OpenAI’s whole thing is to make people use AI at work.Some AI models already exhibit signs of “model collapse,” which is where AI models further trained on AI-generated text can become worse and worse. In this case, some of the people hired to partially stop that happening are themselves using AI-generated responses to train OpenAI’s models.One contractor said they see people using AI “all the time and people are let go for it all the time, it’s pretty much the one thing that will get you kicked off ASAP.” The person said, “in a group of thousands there are tons that have been caught.” 💡Do you work as a prompt reviewer for OpenAI, Anthropic, or another AI company? I would love to hear from you. Using a non-work device, you can message me securely on Signal at joseph.404 or send me an email at joseph@404media.co.Last week, 404 Media revealed Project Lily, in which OpenAI has hundreds of contractors reading real ChatGPT users’ prompts and conversations which can include personal information. Those contractors then rate and critique the responses ChatGPT generated, including making sure that the responses are not too sycophantic or anthropomorphize ChatGPT. That reporting was based on internal documents and conversations with a person who works on the prompts. 404 Media has now obtained other internal documents and spoken to three contractors doing work for OpenAI across various projects. Those projects can include more than ten thousand contractors, according to one of the internal documents.One of those documents says contractors must not use AI themselves for their work. “Do not use AI detection tools, or AI yourself,” one document describing the work of contractors who are hired to review the work of other contractors, including catching them for using AI, says. “Do not use GPTZero or any other AI detection tool. They are not reliable. Reviewers may not use AI either, including Grammarly and AI translation, to review, write feedback, or write comments.”The document continues, “Do not tell evaluators why you suspect AI. It is easier for them to hide if they know what you look for. Judge the overall pattern, not one clue.” All three of the contractors said reviewers are told not to use AI in their work. Two of the sources said people have been fired or offboarded for using AI. 404 Media granted the contractors anonymity as they weren’t permitted to speak to the press.The contractors who review other contractors’ work are told to be on the look out for tell-tale signs of AI use. That can include repetitive words, AI-style punctuation — which might include over zealous use of the em dash — and contractors finishing their work very quickly.In related Slack channels where people ask each other for advice, a lot of people will post an example with the question, ‘Is this AI?,’ one contractor said.“Usually the answer is yes,” the person said.One contractor said they used AI while helping to train OpenAI’s models and shared what they presented as their termination letter. It said their employer had identified issues with the “authenticity” of their work. “I’m not a bad person or worker. I just needed a little boost and turned to AI to help me which eventually led to my downfall,” the contractor told 404 Media. “I felt no joy in the work or that I was contributing to society in any way.”Two of the contractors 404 Media spoke to worked for Mercor, an AI-training company that hires the contractors who in turn review ChatGPT-related material. A Mercor spokesperson told 404 Media in a statement: “Our experts are hired for their expertise and judgement, which is essential to the ongoing advancement of AI. Our contracts strictly prohibit the use of LLMs to complete projects and we enforce that. We invest heavily in our tools and systems to detect misuse and ensure our experts comply with project rules and contract terms. When we confirm an expert has used AI to complete a task, we immediately remove them from the project.”404 Media spoke to a fourth contractor who has worked on training models for various AI companies. They said they sometimes purposefully chose the worst responses because they wanted to actively sabotage the models’ training. “I did feel guilty about doing this kind of work at the start,” they said. “I either pay zero attention to the results and choose randomly or purposely choose the [worst] output. I’m not sure how much of a difference it actually makes since there are hundreds of other people also rating prompt results, but it does feel like I’m getting paid to make AI worse.”OpenAI declined to comment on its contractors being fired for using AI. About the author More from Joseph Cox More like this OpenAI Catches Up to AI Market Reality: People Are Horny Just two months ago, Sam Altman acknowledged that putting a “sex bot avatar” in ChatGPT would be a move to “juice growth.” Something the company had been tempted to do, he said, but had resisted. Samantha Cole · Oct 21, 2025 People Are Crashing Out Over Sora 2’s New Guardrails OpenAI’s Sora 2 platform started just one week ago as an AI-generated copyright infringement free-for-all. Now, people say they’re struggling to generate anything without being hit with a violation error. Samantha Cole · Oct 8, 2025 OpenAI’s Sora 2 Copyright Infringement Machine Features Nazi SpongeBobs and Criminal Pikachus The main use of Sora appears to generate brainrot of major beloved copyrighted characters, to say nothing of the millions of articles, images, and videos OpenAI has scraped. Jason Koebler · Oct 1, 2025 How AI Chatbots Are 'Deskilling' Human Empathy Sherry Turkle's latest book, Artificial Intimacy: Who We Become When We Talk to Machines, comes out on September 29. Samantha Cole · Sep 21, 2026 Is Your City Using Axon License Plate Cameras? We Need Your Help 404 Media is filing public records requests around the country to find out how cops are using Axon's ALPRs. Here is how you can do that too. Joseph Cox · Sep 21, 2026 ‘Supermountains’ Buried Under Antarctica Fueled Explosion of Life, Scientists Discover Some 600 million years ago, the supercontinent Gondwana produced a colossal mountain range that helped set the stage for all modern complex life, according to a new examination of tiny ancient rocks. Becky Ferreira · Sep 19, 2026 Advertisement Go ad free Advertisement Go ad free • Hide Unparalleled access to hidden worlds both online and IRL. 404 Media is an independent media company founded by technology journalists Jason Koebler, Emanuel Maiberg, Samantha Cole, and Joseph Cox. About Twitter Join the newsletter to get the latest updates.
Success Great! Check your inbox and click the link to confirm your subscription © 2026 404 Media. Published with Ghost. |
OpenAI employs thousands of contractors to improve its AI models by having them read real ChatGPT user prompts and other data, intending to inject a human element into the model’s responses. However, this system has been complicated by the fact that multiple contractors have been terminated for using artificial intelligence to train the very models they were hired to support, creating a significant irony within the industry. This situation arises in the context of concerns regarding model collapse, where training AI models on AI-generated text can lead to further degradation of the model quality. Some personnel involved in mitigating this collapse were themselves engaging in the use of AI-generated responses for training purposes. In an attempt to maintain quality and integrity, internal guidelines were established for these contractors. One internal document explicitly instructed contractors not to use AI for their work, and further specified that they must refrain from using AI detection tools, such as GPTZero, or other tools like Grammarly or AI translation when reviewing, writing feedback, or creating comments. The guidance emphasized judging the overall pattern of work rather than isolating single clues, noting that reviewers should watch for patterns like repetitive wording, specific stylistic punctuation, or unusually fast completion times as indicators of potential AI use. Contractors who review the work of other contractors were explicitly warned not to use AI, and they were instructed not to reveal to evaluators why they suspected AI use, as this provided reviewers with an easier opportunity to conceal their findings. This oversight necessitated a communication channel where contractors could share examples, such as posing questions like “Is this AI?”, within internal Slack channels. The experience of these contractors varied, and some faced termination for their misuse of AI. One individual shared that they relied on AI to gain a boost in their work, which ultimately led to their dismissal, expressing a sense that they had contributed nothing meaningful to the work. A separate AI training company, Mercor, reinforced these strict stipulations, asserting that their experts are hired based on their expertise and judgment, and that contracts strictly prohibit the use of large language models to complete projects. Mercor stated that they invest in detection systems and immediately remove experts who misuse AI, confirming that confirmed instances of AI usage result in immediate removal from the project. Furthermore, one contractor recounted intentionally selecting the worst possible responses during training, a behavior which caused moral conflict because they felt they were being paid to actively degrade the models’ performance. Although OpenAI declined to comment on these employment actions, the situation highlights the tension between utilizing AI for development and maintaining the authenticity and integrity of the human expertise involved in training advanced systems. |