Who has used as argument that passkeys are phishable?

The argument is that passkeys include "attestation" as part of the official standard, which is a method explicitly intended for services to be able to securely identify what local authenticator you used with the explicit purpose of allowing the service to reject access if it's not considered "safe". Can you say that about TOTP?

The problem is not the general idea of passkeys.. there are actually good things in the standard. The criticism was that there's also a part in the standard that seems perfectly designed for companies to use them in anti-consumer practices.