OpenAI Wants Contractors to Share Their Old Work Files

So here's something that caught my attention today. OpenAI is apparently asking its contractors to upload actual work files from their previous jobs. Not examples or sanitized versions. The real deal.

I've been following AI training data controversies for a while now, and this one feels different. We're talking about contractors being asked to share potentially confidential work materials from past employers. An intellectual property lawyer quoted in the original report didn't mince words, saying OpenAI is "putting itself at great risk" with this approach. And honestly? They're probably right.

Think about what this means. If you're a contractor who worked on, say, marketing copy for a tech startup last year, OpenAI might be asking you to upload those exact files. Or if you designed internal training materials for a corporation. That stuff wasn't meant for public consumption, let alone AI training data.

The timing is interesting too. We're in 2026, and OpenAI has been under increasing scrutiny about where they get their training data. Just last month, several publishers filed complaints about unauthorized use of their content. Now this? It feels like they're trying to find new sources of high-quality, real-world data, but wow, talk about opening a can of worms.

What really gets me is the position this puts contractors in. Do you risk your professional reputation by sharing client work? Or do you say no and potentially lose future OpenAI contracts? Not exactly a great spot to be in. I'm curious to see how this plays out, especially if any former employers catch wind of their proprietary materials ending up in AI training sets.