Pitch about bot prevention startup.

I was at a cybersecurity conference today and there was a section for pitching startup ideas.

One of the people that came pitched an idea about a bot detection / whatever you want to call it startup firm. I found a really short description of the technique and if you subtract the buzzwords it was something along the lines of "using computer vision to validate user accounts in the scope of bot detection".

Now I didn't get a chance to talk to the guy since I didn't feel like networking today and thus i couldn't find more about what he wanted to do.
As I see it there are just three options (maybe you could suggest more) :

1) He wants to use computer vision (CV) to validate user accounts once at creation (or periodically), similar to dating apps or whatever.
My take about method 1 : This should be easy to bypass a solid percentage of the time but who knows, I feel like falsifying selfies is achievable and I also saw some people that claimed to be able to do it but I never tried it myself.

2) He wants to use CV to match selfies to id photos
My take : This cannot be bypassed to create bot accounts without falsifying IDs and that might be some type of fraud (I'm not sure, don't quote me).

3) He wants to do a combination of method 1 and analyzing posts of the user, to see if there are multiple selfies / photos of groups of real people that include the user in such a way as to give some kind of hidden rating and once a threshold is reached the account gets flagged as real for a certain amount of time.
Take : This could be tricky to bypass and I feel like it would be the best method with the least inconvenience to real users I guess. Now AI/ML is out of my league and I don't know how much infrastructure you would need to do this reliably but if possible I would go for this implementation.

Is it just a cash grab ? Is it feasible ? If it is, could people bypass it ?
What do you think ?

Also, if I go to the convention tomorrow and I feel like talking to the guy, to which board do you feel like I should do an update ? Black Hat SEO or just keep it in the lounge ?

cyberpimp said:

1) He wants to use computer vision (CV) to validate user accounts once at creation (or periodically), similar to dating apps or whatever.
My take about method 1 : This should be easy to bypass a solid percentage of the time but who knows, I feel like falsifying selfies is achievable and I also saw some people that claimed to be able to do it but I never tried it myself.

2) He wants to use CV to match selfies to id photos
My take : This cannot be bypassed to create bot accounts without falsifying IDs and that might be some type of fraud (I'm not sure, don't quote me).

3) He wants to do a combination of method 1 and analyzing posts of the user, to see if there are multiple selfies / photos of groups of real people that include the user in such a way as to give some kind of hidden rating and once a threshold is reached the account gets flagged as real for a certain amount of time.
Take : This could be tricky to bypass and I feel like it would be the best method with the least inconvenience to real users I guess. Now AI/ML is out of my league and I don't know how much infrastructure you would need to do this reliably but if possible I would go for this implementation.

Click to expand...

Sounds like the kinda thing that platforms like FB will eventually implement. They're already asking users for ID verification, so this isn't much of a stretch.