OpenAI Contractors Fired for Using AI to Train the AI
CarissaVeliz · x · 2026-09-24
404 Media reports that multiple contractors hired by OpenAI to read real ChatGPT prompts and improve model responses have been fired for using AI to do their AI-training work.
- One contractor says such use is rampant: "in a group of thousands there are tons that have been caught," and using AI is essentially the one offense that gets you removed ASAP.
- The irony: AI training companies fire people for using AI while OpenAI's whole business is making people use AI at work.
- The piece notes some models already show signs of "model collapse" from training on AI-generated text — and the humans meant to prevent that are themselves feeding AI-generated responses into training.
More from Companies & People
- Tracking global AI talent flow: the 'China makes 50% of AI talent' claim examined — kevinsxu · 2026-09-24
- Who Speaks for AI Companies to Developers? Anthropic's Missing Frontman, Karpathy Tweeting in Slack — dejavucoder · 2026-09-24
- Linear CEO: Consumer AI hardware is skipping the nerd-early adopter stage too fast — _AustinCalvert_ · 2026-09-24
- Google, OpenAI and Anthropic's new standards body named Standards Authority for Frontier AI — steph_palazzolo · 2026-09-24
- Baidu's Apollo Go robotaxis have tested on London streets for over a month — Baidu_Inc · 2026-09-24
- Max Planck's ERC-funded BrainAlign project opens two postdoc slots on brain-aligned LLMs — mtoneva1 · 2026-09-24