I hate how legal has forced every trust & safety team to just blanket reply "You were banned. We won't tell you why. We won't overturn. Go away." It's absolutely impossible to contest without public attention or legal action, and is often just a simple mistake.
More often than not it's not a mistake.
Very simply put, fighting abuse is very asymmetrical and a lot of the approaches just even the playfield. They will make you put in more effort for them to put in effort.
Unfortunate, but the only way it's sustainable.
So? A false positive rate of 50% is wholey unacceptable when banning people from critical online infrastructure.
> Unfortunate, but the only way it's sustainable.
This common talking point is just BS apologetics.
There is a long history of ways to manage disputes without sending such useless explanations. These companies are just to cheap or lazy to bother.
Where'd you get that number?
> There is a long history of ways to manage disputes without sending such useless explanations.
Sure, the legal system.
Your use of the phrase "more often than not" applies up to 50% and is thus meaningless given that such a false positive rate is unacceptable.
> Sure, the legal system.
It's pretty hard to file a legal claim without facts to contest...which is precisely the point.
Businesses throughout history have been sustainable without needing to ban people at random with AI. This AI hell is a new phenomenon in the past 10 years and it's purely driven by cost-cutting.
When the cost of abuse lowers, so must the cost of defense. So of course cost-cutting has to happen. Otherwise it's going to be a financial DoS.
All that is though only if you want services to have free tiers. If you're willing to pay for everything then sure, it becomes easier. I'm actually quite certain one will encounter much less "AI" when stuff is actually being paid for.
I guess besides free users, the other exception is payment processors. I don't envy Stripe or PayPal.