A sensible approach might be to outsource the KYC process to a central state agency. We already have rudimentary forms of that-- like the fingerprint-clearance cards you sometimes need for working with the vulnerable. Why would it be any harder to grep the database for financial criminals than sex offenders?

Give me the ability to generate a signed certificate that says "hakfoo is not currently denylisted for financial crimes" and let me submit it as part of account onboarding. Maybe include some new, non-SSN, low-stakes identifier so I can't make forty different certificates and pretend to be forty people.

Such an agency can be more privacy forward than any private player: there's no financial motivation for them to be "sharing data with our partners and affiliates", and is structurally disinterested in expanding its reach into other data-grab products (looking at you, credit reporting services).

If the banks and such really have to do some sort of data-derived KYC-theatre-- looking for structuring and spurious patterns-- it could still be done with minimal personal data-- tracking the accounts by UUIDs.

This would certainly help with reducing the structural incentives to sell the data, and it might help with preventing its theft. But it might instead make the theft problem worse, because the database it created would be both larger and more comprehensive and reliable. Such a database would be a more valuable prize, whether for a military coup, an invasion, employee corruption, or simply an infosec failure like this one. The reduction in attack surface and improvements in defender competence enabled by such centralization might not be enough to compensate.

See Stuxnet for one such example—it wasn't related to privacy, but it seems unlikely that the Iran's National Organization For Civil Registration is any better able to prevent foreign penetration of its systems than its nuclear energy agency was, so presumably the Mossad has a complete and up-to-date copy of its national identity database. Such information seems likely to be very valuable in circumstances such as the assassination of Ismail Haniyeh and the next few months of open war between Israel and Iran.

I suspect the databases useful for KYC already exist. The denylists that have to be checked against, for example.

There might be a new record of inquiries or certificates generated, perhaps, but that doesn't really need to be logged.

Identity is always an ugly technical problem, and IMO the federal option has the least ugliness. Yeah, you have one database, but that also means you have only one place to monitor, keep up to date with state-of-the-art security norms, and one giant legal hammer to go after anyone who breaches it.

* Devolving the problem to lower-level governments just means you replace one database with dozens, and get to deal with varying security and use paradigms, and interoperability and data synch that's clumsy at best. How long can you keep a scam running using Jane Doe in Idaho's credentials until the system is synched and learns she actually moved to West Virginia?

* Using private entities to manage it would have all the worst incentives for misuse and abuse, and either gives one bid-winner the keys to the economy or creates a bunch of fiefdoms incentivized to limit interoperability.

* Any sort of bottom-up "self-attestation" or web-of-trust sort of solution would be difficult to bootstrap and require generating mountains of new legal precedent and tooling to support.

They exist so the obscenely wealthy can have financial privacy via shell companies, and the non-obscenely wealthy can be financially surveilled.