Classical Public Key Infrastructure (PKI) provides static identity guarantees — it provesthat a public key belongs to an entity, but cannot guarantee that a given transactionoriginates from that entity at that moment. This gap enables replay attacks, key reusevulnerabilities, and long-term compromise cascades.EIDA (Ephemeral Identity Distribution Architecture) addresses this gap by introducingidentity-level forward secrecy. The core primitive, ePriPub, is a single-use ephemeral keypair derived per transaction from the user’s permanent identity, which never leaves ahardware-backed isolation zone. Once used, the ePriPub components are destroyed andthe CA issues a fresh single-use certificate — no persistent state is required.EIDA shifts the trust model from centralized authority verification to cryptographictransaction uniqueness, distributing security to the user without requiring trustedintermediaries.Today, companies responsible for securing user identities are under constant attack. Asingle breach of a centralized security system can expose the personal data of millions ofusers. This is an inevitable consequence of the current architecture: security isconcentrated, and concentration creates high-value targets.EIDA proposes distributing this responsibility to user devices. In a decentralized model,an attacker who compromises a single node gains access to only one person’s data — notmillions. Large technology companies are no longer the primary target. The attacksurface shrinks from a single critical system to millions of individual devices.Classical PKI reinforces this centralization problem. It guarantees that a public keybelongs to an entity, but cannot verify that a specific transaction originates from thatentity at that moment — enabling replay attacks, key reuse, and long-term compromisecascades.The fundamental vulnerability of current internet security architecture is concentration. This is not an implementation failure — it is an architectural one. When identity verification depends on centralized systems, those systems become high-value targets. The attack surface shrinks only when the architecture itself is distributed.
Güven Acar (Sun,) studied this question.