No but if they forget to strip those before training the models, it’s gonna start spitting out licenses everywhere, making it annoying for AI companies.

It’s so easily fixed with a simple regex though, it’s not that useful. But poisoning the data is theoretically possible.