If I clone/fork that repo before the license change, and start putting any amount of time into developing my own fork in good faith, they shouldn't be allowed to claim a clerical error when they lied to me upon delivery about what I was allowed to do with the code.
Licenses are important. If you are going to expose your code to the world, make sure it has the right license. If you publish your code with the wrong license, you shouldn't be allowed to take it back. Not for an organization of this size that is going to see a new repo cloned thousands of times upon release.
For the same reason you cannot publish a private corporate repo with an MIT license and then have other people claim in “good faith” to be using it.
All they need is to assert that the license was published in error, or that the person publishing it did not have the authority to publish it.
You can’t “magically” make a license stick by putting it in a repo, any more than putting a “name here” sticker on someone’s car and then claiming to own it.
The license file in the repo is simply the notice of the license.
It does not indicate a binding legal agreement.
You of course, can challenge it in court, and ianal, but I assure you, there is president in incorrectly labelled repos removing and changing their licenses.
Licenses are important. If you are going to expose your code to the world, make sure it has the right license. If you publish your code with the wrong license, you shouldn't be allowed to take it back. Not for an organization of this size that is going to see a new repo cloned thousands of times upon release.