A single patent tells you what one team built. A patent family tells you what a company has decided to defend. American Express's US12,093,948, granted September 17, 2024, is interesting less for any one limitation than for being the third issued grant in a tight lineage — and that lineage is the portfolio story worth telling.

The shared specification across the family describes a recognizable architecture. A zero-knowledge-proof purchase system comprises a customer device, a merchant system, an issuer system, and a blockchain network carrying a ZKP smart contract; the system runs a zero-knowledge proof algorithm with three named pieces — a key-generator function, a proof function, and a validate function — to enable purchases where the buyer proves something true (that they are authorized, that a condition holds) without exposing the underlying identifying data.

A zero knowledge proof purchase system and process using blockchain is disclosed. The system may comprise a customer device, a merchant system, an issuer system, and a blockchain network having a zero-knowledge proof (ZKP) smart contract.

Now the landscape view. The same abstract — nearly verbatim — anchors three grants: US11,151,558 (issued October 2021), US11,748,750 (September 2023), and now US12,093,948 (September 2024). The same inventor team appears throughout: Andras Ferenczi, Dallas Gale, Nilesh Jadhav, and Harish Naik. Three grants, one specification, escalating issue dates, consistent inventors — that is the fingerprint of a continuation strategy, where an assignee files successive applications off a common disclosure to capture claim scope from multiple angles and keep a live application pending as the technology and the competitive field move.

The CPC clustering tightens the read. The grants sit across G06Q 20/389 (payment protocols / transactions via networks), G06Q 20/3827 and G06Q 20/3829 (security-of-transaction and secure-key handling for e-payments), H04L 9/3221 (the zero-knowledge-protocol class), H04L 9/0643 (hashing), and — on the later grants — H04L 9/50, the consolidated blockchain class. Watch that last detail: the original 2021 grant carried H04L 2209/38, the older blockchain tag, while the 2023 and 2024 grants carry H04L 9/50. The migration across the family is a clean illustration of why counting an estate by a single CPC code undercounts it; the same invention lineage straddles two blockchain classifications.

The assignee identity is the part that should make a landscape reader sit up. This is not a crypto startup; it is a card-network incumbent. American Express building a privacy-preserving, blockchain-settled purchase flow — and fencing it across three grants — is a data point about where the established payments rails think on-chain settlement is going. Set it beside the broader field: the zero-knowledge-payments space is crowded with IBM, nChain, Visa, and others, but the incumbents staking continuation families are the ones signaling durable intent rather than experimentation.

What this Grant Watch is not claiming: that Amex has cornered zero-knowledge payments. The field is deep, and these grants are bounded by their specific customer-merchant-issuer-ZKP-contract architecture with the recited key-generator/proof/validate functions. A different privacy-preserving payment design — a rollup-settled stablecoin transfer, a confidential-asset scheme, an identity-credential proof — does not read on this family merely by using a zero-knowledge proof. The point is narrower and more useful for mapping the terrain: a major card network has methodically built and maintained a three-grant estate around one on-chain private-purchase flow, and the continuation pattern says it intends to keep extending it.

It is worth being explicit about how to read a continuation family without overstating it, because the temptation runs both ways. A skeptic might wave off three grants as one invention counted thrice, which undersells it — successive grants off a shared disclosure typically carry different claim sets, so the family can cover the same architecture from method, system, and apparatus angles, broadening the practical fence even when the specification is identical. An enthusiast might read three grants as proof of dominance, which oversells it — the claims are still bounded by the recited customer-merchant-issuer-ZKP-contract structure, and the field around them is dense. The disciplined middle reading is that the family signals durable commitment to one flow, with claim coverage thickened by prosecution, and nothing more.

The named functions are also a useful scope anchor for anyone comparing this estate to competitors. The algorithm is recited in terms of a key-generator function, a proof function, and a validate function — the canonical three-part shape of a zk-SNARK-style scheme (setup, prove, verify). That generality is a double edge: it lets the claims read across whatever specific proof system is plugged in, but it also means the distinguishing work is done by the surrounding payment architecture, not by the cryptography. An accused design that uses a different transaction topology — no issuer system in the loop, or settlement off-chain — has room to operate even while using the same proof primitive. The architecture, not the proof math, is where this family draws its lines.

For anyone tracking which incumbents are fencing what in crypto IP, file this under payments-network privacy. The estate is real, the strategy is legible, and the right way to weigh it is by the family and the prosecution pattern — not by treating the newest grant as a standalone landgrab. All three canonical records are deep-linked above so you can read the claim sets and watch the scope shift grant to grant.