"open source developers will continue using AI models that currently exist to benefit their coding were it can" does not undo that all mainstream, commercial AI agents are the product of theft of labor from all the programmers, artists, authors, etc of all the media that the models were trained on.
You can say the ends justify the means, but you have to own that commercial AI agents, and the work they produce when wielded by humans, are a profit engine for thieves. At the end of the day, those programmers are using an AI agent that is being sold to them by a corporation that is profiting off of theft. "But the AI tool works well!" isn't the end of the moral discussion. How did the AI tool get so good at programming?
Copying a FOSS project for personal use and development isn't unethical. Copying the FOSS project into an AI model and then selling that AI model for profit is. And it's been done on a scale that's difficult to fathom and impossible to undo. Persons whose crafts have been irreparably undermined by these models have had their livelihoods impacted and there will be no means of exacting justice from those who have committed to this technology because they are simply too wealthy to hold to account.
If you're a regular guy and you try to download and disseminate information like Aaron Swartz, suddenly the law has a lot to say about intellectual property theft.