@supergarv Per their own update, that's only training on the things you send to copilot and it generates in return.
It's too late for the git repos you've got on there, they've already used that, unfortunately.
@KayOhtie I thought so too but the wording is really ... general? And who knows what they feel is "interacting" with copilot.
But anyhow. We all know its a free-for-all at this point.
@supergarv
I would assume this only refers to the data involved when you use Copilot, that is, it's irrelevant if you don't use Copilot, but maybe I'm too naïve?
(to be frank I expect them to already be training their models on our code without telling us, let alone give us a way to opt out)
@glocq I thought so too, but then.... it was enabled for me just like that, and I would not have done that myself.
And yes, reality is long ahead of any consent. I just dislike the casual mail I got:
"This approach aligns with established industry practices and will enable our models to deliver more context-aware Al coding assistance. We have tested this with Microsoft interaction data and have seen meaningful improvements, including increased acceptance rates in multiple languages."
> it was enabled for me just like that, and I would not have done that myself.
What I meant is, the setting might have been enabled by default, but only be relevant if you actually use copilot. whatever 🙃
> This approach aligns with established industry practices
« If everyone else adds your work to their plagiarism machine, then it must be ok to add it to ours too »
@supergarv Damn. They had this turned on here too. I guess today's the day I migrate away from GitHub. Luckily I've already got a Codeberg account set up.