I recently noticed that someone compromised my identity and opened a retail credit card with my information. Having worked in retail where I opened store credit cards, I know that employees don't always reliably verify someone's identity, and often times all someone needs is a social security number to open a fraudulent account.<p>But now, with modern Machine Learning applications, are banks deploying software to actually verify that someone's signature is at least 80% legitimate? Or are there any startups/projects/organizations/open-source projects anyone might know that are actively working on this?
If ML was used on my signature I would never be able to use a bank, pretty sure I've never written it remotely similar.<p>All I consider a Signature to be is like a confirmation, pretty much the same as a checkbox or clicking an OK button, just in a more inconvenient form
My previous bank took 4 samples of my signature when I opened an account. When one of them was off, the software could tell that and I'd be asked to sign again, until the computer had 4 "similar enough" signatures.<p>And then, in reality, I think it's only checked against when you someone deposits a high amount check you've given them or when you do some other substantial trnasactions.