I'm kind of wondering where this sentiment is coming from that LLMs are stealing open-source code. Isn't it just the same as someone learning programming by rea

I'm kind of wondering where this sentiment is coming from that LLMs are stealing open-source code. Isn't it just the same as someone learning programming by reading and working on open-source code and then writing closed software with that knowledge, where is the difference? Or is it just closed models that are problematic and would open models like Llama/DeepSeek be acceptable?