Fortunately for Nymbot, its “anonymous mode” and everything about it is verifiable as the app/platform code is open source. By design, it is private and not possible to link messages passed through on the platform to anyone. Then the anon mode goes a step further by using blind vouchers to transfer your credits to throwaway keypairs, that obfuscate it even further.

yes it does fall into private ai category. based on the claims (i havent read the source) it provides anonymization of traffic to the providers, but that addresses a very specific issue - attribution of the conversation to specific identity. it does not protect the privacy of the conversation itself, at least not from the provider. So if you upload sensitive documents or prompt it with releaving data your identity might not be tied to the account, but the data can still get out. thats where the confidential inference and confidential computing part comes in, you can get much better assurances there provided that its implemented right and you can attest to a reproducible stack that is not leaking your data somewhere.