To be fair Microsoft haven’t done as much integration with their eco system; such as integrating it with Azure DevOps as they could. And long may it continue.
Not a fan of defending GitHub, but TBH them launching GithubActions made me abandon all other CI platforms I was using in the past, I even left GitLab just for that.
A 99.5 means over one year you will need to wait about 40 hours total for the issue to be solved. What kind of high pressure crud app are you making where this is a problem?
Oh yes I can totally believe they use public and private repos for their model training. It is surprising they aren’t ahead of the curve for AI coding like Claude code, codex or cursor because of this.
I very much doubt that. Imagine the PR nightmare and potential liability if someone managed to extract from the model credentials in a private repository.
Explain to me how using OPEN source for training is theft? I've (admittedly skimmed) quite a few licenses, but I really dont remember anyone restricting reading/indexing source code?
MIT license includes: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
The argument is that AI models trained on open source should include the copyright info for all projects blended into the weights. And since they don’t, it violates the terms of the license.