Amateurishly solutioning this, reality appears to be (1) information will leak somehow; nothing is air-tight (2) identity would mitigate the majority of common problems. "Identity" would be a unique identifier on "the chain" for all entities who want to play. Humans get one each. Zillions of bots (linking eventually/elaborately to unique human ids) get one each. Identity registration will have requirements that all terminate at an endpoint that can have its bad behavior mitigated (prosecution, bot-termination), effecting the entities bona fides. So skin is in the game. Entities will be averse to interacting with anything & everything with any malicious intent.
The spinal cord of this solution is "trust". Bad things will be attempted, but through the chain and algorithms, they can be traced & mitigated. And as usual, society makes the cost of crossing the line too bitter for 99.9% of the entities to taste (revoking key trust domains, jail, etc.).