Is it stealing if AI learns from artists' and writers' work without permission?

Legally, it's an open question — courts are still deciding — but ethically, it feels like stealing to many creators because their work was used to build commercial AI systems without consent or payment. Here's what's actually happening. When you use an AI image generator to create art 'in the style of' a living artist, the model didn't study their work the way a human art student would. It was trained on millions of images scraped from the web, and the artist never opted in. A 2023 lawsuit from Getty Images against Stability AI centers on exactly this: Stability allegedly used 12 million Getty photos to train its model without a license. The writers' side is just as heated. The Authors Guild filed a class-action suit against OpenAI in 2023, claiming ChatGPT was trained on pirated books. I think the discomfort comes from the asymmetry. A human painter who studies Picasso for years transforms that influence through their own hand and life experience. An AI model mathematically reconstructs patterns at scale, and the company that built it charges $20 a month. That's a different thing entirely. One practical insight: if you're using AI tools for commercial work, stick to models that license their training data or let creators opt out. Adobe Firefly, for example, trained only on licensed and public domain content. For more on the legal side of AI-generated content, see our guide on AI content copyright and legal issues. **Related**: Can I copyright something I made with AI? | Which AI image generators use ethically sourced training data?