That's why I said "I think", it's not really clear. That being said:
> Will CodeWhisperer produce code that looks similar to its training data
> If CodeWhisperer detects that its output matches particular open-source training data, the built-in reference tracker will notify you with a reference to the license type (for example, MIT or Apache) and a URL for the open-source project.
They seem to at least have some protections in place to prevent CodeWhisperer from spitting out existing code without attribution as shown in the Twitter thread. They also only mention MIT and Apache.
