In the UK there is a big push around "open banking"[1] which will bring this into the 21st century and allow for proper programmatic access to data. It's still in it's infancy but the sector here is transforming around it.
[1]: (https://www.openbanking.org.uk/customers/what-is-open-bankin...)
https://www.mccarthy.ca/en/insights/blogs/cyberlex/open-bank...
"The ACCC envisages that the first tranche of open banking rules will come into effect from 1 July 2019. Within 12 months, all Australian banks, including the related brands of the big four, will be brought within the scope of open banking."
Basically, my password is hashed to see if I can log on. Then it's passed through a PBKDF to get the decryption key for my actual accounts, then that information gets sent to the scrapers to do the actual job. They don't store the keys after the job is done. The upshot is that a full database breach doesn't result in any bank credentials leaking, at the cost of inability to update accounts without the user explicitly logging in.
In order to do so, they most likely keep your banking password around in memory.
Note: Mint uses OAuth for access to Chase bank accounts, which is great. Last I checked Plaid does not.
At some point I think it's on the banks to offer OAuth APIs, then Plaid can swap out one-by-one (if it hasn't already started).
Nylas is facing the same challenge in the email space. They have oauth for gmail, but user/pass for Exchange/SMTP.
So your username and password are just kept in some internal database somewhere? The scrapers probably decrypt the credentials in-memory.
Also - "scraping" data off of an undocumented API sounds risky. How can I guarantee a "scraper" won't accidentally mess something up for me?
Plaid's implementation was aggressive (screen scraping, etc) but many banks are blocking that now and some, like JPMorgan Chase, have created APIs based on OFX (the industry consortium for secure data exchange) to allow controlled data access.
You, the customer, should always be able to choose which targets get to receive your data. Via OAuth mechanisms you grant them access without sharing your login/pw, and you can revoke at will.
It's worse than that. Sharing passwords with third-parties typically "voids the warranty." If money is stolen from your account, the bank can deny reimbursing you because giving a third-party your password voids their zero-liability guarant guarantee.
Banks in the US are really behind the ball on APIs, it's true. Just recently things have started to change though. I'm the cofounder of Treasury Prime (https://treasuryprime.com/) and we have a network of banks in the US who offer API access.
It doesn't solve Plaid's use case (getting data out of 1,000s of banks), but it's great if you need deep integration into a single bank, like if you're writing a fintech app for example. If anyone would like access, feel free to email me: [email protected]
Also, big congrats to Plaid on the fundraise!
I'm not sure how to feel about this, because I understand that banks' lack of open API access is the central problem. But it seems irresponsible to present Plaid as a secure solution, when its login system is technically a phishing page.
I think a much cooler, probably safer, solution would be a mobile SDK that runs the scrapers directly from the user's phone, instead of on Plaid's servers.