This is dangerously wrong coming at least a decade after there are entire communities devoted to unmasking performers’ real identities and multiple reverse image search tools exist as apparent businesses. That used to be a human-driven practice - I first heard about it coverage of the Chinese internet mobs from the perspective of victims of misidentification - but like everything else it’s reportedly adopting AI. Here’s a story which got a bit of discussion a few years back:
https://thenextweb.com/news/creepy-programmer-builds-ai-algo...
One of the big things to remember is that these systems don’t need to be perfect, or even close, to cause harm. Even if they were only 10% accurate, that’s still a lot of people living with the question of whether the person they just met knows or whether today is the day some nut sent those links to HR. You can’t rely on getting lost in the crowd any more.
And more importantly, said creeps would be the one who would have an inappropriate behavior in the workplace regardless of the tools they have at their disposition.
It’s also severely optimistic to think that the guy doing it will suffer the consequences: if you search the news, you’ll find plenty of examples of cases where someone thought they knew the attacker but wasn’t able to prove it. Moreover even if they could prove it and the attacker did suffer consequences, it won’t magically wipe everyone else’s memories.