We're a bit light on detail here but it's worrying that it's 2025 and Google isn't flagging "looks like" @google.com messages.
I'm assuming this is a dirty unicode hack and not something worse: no DKIM or an actually compromised sender.
The whole thing stinks.