The Counterfeiting Century

For ten thousand years we trusted objects because we trusted the people who remembered them. That just broke.

Let me start with something simple, then something obvious. Between the two sits the work of the next decade.

The simple thing: for ten thousand years, humans trusted objects because we trusted the people who remembered them. A sword was real because the blacksmith vouched for it, or the king who carried it, or the monastery that kept the record. Trust was local, because memory was local. The archive, the guild, the museum, the manufacturer's heritage department — these are the original trust infrastructure of civilization.

The obvious thing: we are entering a world in which a machine can generate, for almost nothing, a perfectly convincing image of a sword that never existed. A letter from a king who never signed it. A monastery record that was never written. A video call with five people, none of whom are real. In early 2024, a finance worker in Hong Kong transferred twenty-five million dollars after exactly such a call. This is not speculative. It is this morning's commodity.

The scarce good is no longer generation.

Generative AI is not, at its heart, a content technology. It is a counterfeiting technology. And it is being released into an economy whose entire secondary market — luxury, art, heritage, collectibles, the data we train the models on — still runs on authentication machinery built for the nineteenth century.

When anything can be faked, verification becomes the scarcest good. Not generation. Verification. So the question worth asking is a simple one: who will own the layer that tells you what's real?

There are two ways to build it.

The first is to scrape the world's objects, build an adversarial database, and sell verdicts. That model scales to handbags. It does not scale to watches, to art, to manuscripts — to the things that actually hold the value. Because for those, ground truth doesn't live in a scraped database. It lives in the maker's own archive. It lives in the regimental foundation. It lives in a hundred thousand museums and heritage departments that have quietly accumulated authoritative records for a century or more.

The second way — the only one that scales — is to connect those institutions into one shared index, and pay them for it. That is the provenance network. That is what we are building.

A forger can copy the object. Not the record.

What a forger can copy is the object. What they cannot get into is the record — the accession ledger, the production book, the regimental file, written down decades before, held by an institution. Connect those records, cross-linked across many independent sources, and you have an index of what's real: every claim sourced, every connection verified, every attribution traced to a real object in a real institution.

The institution's raw data never leaves its walls. Our agents query back. We enrich, we verify, we never extract. And most of every dollar a verified answer earns flows back to the institution whose records produced it. We are not taking from the institutions of memory — we are switching on an asset they have never been able to activate.

The moat is not the technology. It's the alliance.

Every institution on the index makes it more valuable to everyone asking is this real? — and every question makes the index more valuable to every institution. No competitor can go back in time and build a fifty-year relationship with a regimental foundation, or scrape a maker's archive without being sued. Time is the moat, by construction.

We are building the verification layer in alliance with the institutions of memory — not in opposition to them.

History has done this before.

Every major technology shift produces an adjacent trust infrastructure that captures more value than the shift itself. The printing press produced the publishing house. Industrial manufacturing produced the trademark. The internet produced the certificate authority and the search engine. Artificial intelligence will produce the verification layer for physical truth.

That layer is going to be built. The only question is who builds it, and on what principles — in the decade when holding something in your hand stops being evidence that it is real.