This is a very confusing read. It gives the impression that the attacker managed to manipulate the email body to insert their phishing link, by talking at length about how the sites.google.com link is suspicious (of course it is, no doubt about that). But at the same time, they don’t say or show evidence that the body was manipulated; in fact quite the opposite.

My understanding is that the DKIM signature contains a bh= field with a hash of the email body. While you can technically also include an optional I= field to limit the body length for hashing, so that an attacker can append to the body, which is a pretty big security hole, it’s probably never used by Google for such short emails (I checked some of my own emails from no-reply@accounts.google.com and they certainly don’t have I=). Therefore to pass DKIM and DMARC the body had to be intact, so the “phishing link” was actually from Google, just intended for a different recipient.

If my analysis is correct then TFA really is a lot of words to say a scary email was forwarded to wrong people to scare them. Scary of course, but much less scary than the “DKIM replay attack” title implies to technical people who are not deep into this subject.

Edit: Oh, I thought “The Takeaway?” was the end of TFA since it had CTA for their product. Apparently there’s an update below explaining the link was actually part of a Google OAuth app name which was then inserted into Google’s email template. Terrible writing and structuring of the article, burying arguably the most important part of the attack that made it somewhat convincing, and misleading readers to believe the attack can be used to send arbitrary content.

Edit 2: Other commenters pointed out that the screenshot of the email is conveniently cut off so the fixed part of the Google email template isn't shown. The attack is probably even more clumsy then it seems from the quite deceptive crop.