[anti-abuse-wg] AI used for spam and abuse at Microsoft?

Hi,

I have seen a marked increase in email spam abuse from @outlook.com and
even when reported it seems as if Microsoft is powerless (or does not
want to) stop the payday, viagra and other rubbish there users are
apparently sending. I am sure that I am not the only abuse admin that has noticed this and
I wondered if anyone knows what happened or what changed @microsoft
(or am I just so special that they target only my networks?  :)  )

My day today:
I had the pleasure of learning that Microsoft has hard blocked one of
my transactional servers on ALL of their domains and across their
entire platform even Internationally/Everywhere... (IPv4: 5.9.29.165)

http://multirbl.valli.org/lookup/5.9.29.165.html

Of course the domain as well as the IP is completely clean as it does
not send spam, or even bulk email (it is not even an email server, it
is a transactional server)

I also have not received any abuse reports over the past ten years,
from anyone, anywhere.

So, I needed to "apply" to Microsoft to reconsider blocking my IP
number: https://go.microsoft.com/fwlink/?LinkID=614866

Which is a web form, and reminded me of the recent list discussion
about how impossible some webforms are to complete....(Maybe these
forms are just as challenging for AI to complete as for people?)

Never the less, after 20 minutes, I managed to get the form completed
and submitted, only to be quickly informed, via email, that Microsoft
has investigated the issue and that they will not lift the block.

The response was so very quick... Am I on an International blackball
list or somesuch? - Probably not as I am and have never abused anything
or anyone. So, the only other thing is maybe AI?

So, does Microsoft have actual humans lifting blocks? or is it all
automated (and completely broken)?

Abuse and Internet abuse is a very relative thing. Like truth, facts and reality. 

Already we are creating a world where truth and facts are what corporations 
and powerful people decide they are, I am not sure that abuse handling
is something that can already be fully automated in 2019?

Andre