Not sure that is about automation, as much as your banking system has a flaw they need to fix. You can't do that here in Australia. Well... you could, but you can't, not automatically. You would have to identify yourself over the phone, AND, if you separate from someone, you just change your banking status from married to divorced, and there is no way they're allowing anyone to call and make changes to anything, if you can't can't pass all the hoops they throw at you. Here, they will even shift you to their app, which will take a picture of you so they can ascertain whether you are the person or not, all in real-time. I think maybe your system has flaws, which I'm a little shocked they haven't patched something or made something available so that nobody but you can make changes.
Test that theory. Grab a new device, and log into your bank account (or have N do so), with “forgot login / password”. Ditto, go to any university website, and find the tab to add/drop classes… with “forgot login / password”. If you HAVE someone’s personal identifiers? Who has an account? Voila. It’s no small part of why identity theft is a billion dollar industry. Not just credit cards, but the whole shebang. And, at least in the US, it’s virtually impossible to police. As a single incident/case is hundreds of man hours, so it’s only worth it to law enforcement to go after organized crime, hackers, etc. doing thousands of people (child porn & bank fraud & organized crime are where tech resources are focused, and it’s a finger in the dam)… not a “domestic squabble” or targeted stalking of a single person. It simply doesn’t make sense, pragmatically, which I DO understand. Corporations fund their own investigations & simply present all their evidence to police.
AI is ending the automization. By creating a failsafe, in the programming, allowing accounts to be flagged and put into a separate queue requiring old fashioned, in person, face to face verification. Nothing at a distance. Not over the phones, or online, but physical presence. AND? Will “reset” your status to its prior standing when fraud is attested, rather than the person being shit outta luck even WITH a court victory (2 years later), much less with ‘just’ the police report filed (that will never be investigated). Modernly, “all” one has to do is say “I didn’t drop that class, or buy that boat in Amsterdam, or have my appendix out” and the assumption is fraud/indentity theft. Rather than crazy-person trying to get one over on us, f*ck off. But even when one WAS believed? 5, 10, 15 years ago? There was nothing to be done. You’ve lost your financial aid, you cannot see your doctor until you pay for the surgery you didn’t get, etc. etc. etc.
The fail safe is partly a Skynet-Paranoia, wanting to give human operators more power over the machine, and partly because it’s so damn easy to clone anyone’s phone/device, and have their fingerprints/face/account logins & passwords… so identity theft and related fraud have been increasing astronomically over the past decade. When it was done to me, it was restricted to the tiny circle of a few hundred well placed people & elite hackers. Now? Any reasonably tech savvy 12yo can clone a device, mirror, &/or deploy a keylogger. Or use a voice modulator to make a dude sound like a laaaady

. Of course, the kids do it for kicks & bullying. The grownups …for not so dissimilar motives?
Face to Face in person verification 100%? Pfft. Of course not. But fake IDs are usually restricted to kids drinking & espionage. As thieves & the ilk resist their real face attached to anything.
ETA… 10+ years ago my BANK did have a “divorce” queue… which removed all personal identifiers as valid authorizations. As spouses generally know each others. And reset to randomized keys, that were only acquired off-line, in multiple formats. Now? It’s not -yet- universal, but AI being deployed is swiftly making it so.