github.com FAQ
What is the difference between Github Enterprise Cloud (github.com) and Github Enterprise Server (github.umn.edu)?#
Github Enterprise Server is hosted on prem and managed by the SRE team. Github Enterprise Cloud is hosted on github.com as an "Enterprise", which allows for an SSO barrier aroud Orgs and Repos in the Enterprise on github.com itself. DevEx manages Github Enterprise Cloud.
Why do I need a github.com account? What about SSO?#
A github.com user account is required to interact with and commit code to github.com. Your user account on github.com will be linked to the UMN SSO in order to access the Enterprise. This will result in two logins, one for your github.com user and one for interacting with UMN Enterprise orgs and repositories. This will also mean two separate 2 factor authentication setups, One for github.com and another for UMN's Duo. See User Management for more information on accounts.
What about Github Copilot?#
Use of Github Copilot is currently turned off while policy is being discussed.
What about Github Packages?#
Packages is available but currently set to a 50GB limit over the entire enterprise as we investigate usage.
How do I add someone to my Org?#
Org owners can email devex@umn.edu
with your org and the internetid of the person you'd like to add to open a ticket.
How private is code in a private repository?#
GitHub has answered this question in their privacy statement.
The short of it is that they do not access private code except when they have to, or if you have a service turned on that will scan private code (e.g. secrets scanning in the advanced security features). You will be the one to enable such features or if we make any changes across the Enterprise, it will be with notification or collaboration with you because such features are on a per-repo or per-org basis.
As far as whether copilot is trained on content from private repos, GitHub Docs has a comment about use of private code in Copilot Individual, but the comment is not replicated in the Enterprise docs, so it's somewhat inconclusive.
We do not ever recommend allowing user data in code at all since we can only tell you what we are told by GitHub and with AI becoming a lot more ubiquitous, information control is difficult. That said, we know sometimes it cannot be avoided so this is left to your discretion.