[alwaysrunning]

How exactly are they going to stop the NYPD from using this service? Just not register it in their subscription as a service? And does it really matter? If I can use Azure's cognitive search and host my own model that doesn't stop me from using Azure for the same purpose.

[HeatrayEnjoyer]

Using it unofficially through unauthorized workarounds brings doubt into the chain of custody and investigation integrity.

[alwaysrunning]

I'm not sure the courts would consider it a workaround, it's a legit way to host your own model.

https://learn.microsoft.com/en-us/azure/machine-learning/how...

[Maxious]

Clearview AI created an unregulated facial recognition system and let individual cops use it for free as a "trial" Their superiors claim they had no idea https://www.abc.net.au/news/2020-07-13/afp-use-of-facial-rec...

[riffic]

when cops and DAs use parallel construction to build a case, "integrity" is a mere technicality.

https://en.wikipedia.org/wiki/Parallel_construction

[_DeadFred_]

Oh no, they will have to use parallel construction, something they TOTALLY don't have experience doing.

[karaterobot]

You can't stop someone from misusing your technology by updating your terms of service, just like you can't stop a criminal by making their behavior against the law. But I don't hold Microsoft responsible for making the police behave a certain way, or for preventing violations of the rules from happening. As you say, it's just not possible. I'd be satisfied if Microsoft consistently shut down any attempts to use their technology in a systematic way, once they discovered a violation of their terms.