WORM prevents after-the-fact modification, but it isn't very helpful in the case of persistent threats.

The concern is that the tampering has already been committed to the backups. When was the "Break Glass" password last rotated? Is it protected by one or more Yubikeys that were manufactured before they fixed that nasty exploit? What other attack vectors are baked in through malfeasance or human error?

My comment was not in reply to passwords, "yubikeys" or anything else you mentioned, so your techsplaining about those things was a bit misplaced. MY point was that if the backups are on WORM tapes, and we still have those backups, then there's nothing to fear being compromised from those backups. Everything else is outside the scope of my comment.