Essentially because it replicates the way a browser works, similar to youtube-dl. I would very much recommend taking a look through [0] for some more info, but one of the main goals of the project is to avoid use of any Google APIs. This also includes login, as mentioned in an earlier comment[1].

0. https://github.com/omarroth/invidious/issues/130

1. https://news.ycombinator.com/item?id=21545348