Ask HN: Why is it OK for Cursor and Windsurf not to credit their model
Windsurf's SWE-1.5 is really good, and I've heard good things about Cursor's model too.
However there's this nagging feeling I just can't shake: using the base models without credit is not OK.
I know they've done some post-training and a lot of work optimizing the inference, but to use something as substantial as an LLM and not even credit them feels super wrong.
I guess my question is how is this OK, regardless of what the licenses allow. When you fly in an airplane do you care what engines they’re using? Do you care what avionics subsystems are installed and what SoCs they utilize? What makes this situation any different? It’s just that LLMs are the new hotness. That will fade. Why is it OK for AI models to don't credit the training data, even if they reproduce the code almost verbatim? The AI companies are not big fans of crediting - once they get access to _any_ learning data, all attribution disappears. So I am pretty surprised you are even asking this question - of course they won't credit anything they don't have to. This is in their ethics. (Now if the customer requires it, they will expend the effort... I am sure if you pay them enough money, they will credit anyone you want) Why did you not credit the computer, keyboard, internet provider in your post? Outrage over uncredited models... built on uncredited data... Poetic.