On-device AI/ML was big at Google I/O 19. There were at least 4 sessions about inferencing on mobile (TFLite), IoT (Coral) and in browsers.
If on-device ML is real, then Apple and Google have a moral obligation to make harmful content (e.g. porn) detection part of their parental controls as soon as possible.
The real question is why Erdogan took risk an blocked Twitter just 10 days before elections? The only reason I see is this Twitter account with almost 500.000 followers who is revealing government corruption: https://twitter.com/FuatAvni For English version of his tweets see: https://twitter.com/FuatAvniEng (usually translation is delayed 1-2 weeks)
If on-device ML is real, then Apple and Google have a moral obligation to make harmful content (e.g. porn) detection part of their parental controls as soon as possible.