Provenance and authenticity in art, in the AI age

Provenance is one of those words that sounds like it belongs to auction houses and old masters. In fact it is about to matter to almost every working creator, because the problem it was invented to solve — proving where something really came from — has just become everyone’s problem. When imagery can be generated by the million and made to look like anything, the documented story of a work’s origin stops being a connoisseur’s nicety and becomes the thing that anchors trust.

What provenance actually means

Provenance is the recorded history of a work: who created it, when, and the chain of ownership and handling since. In the traditional art world it is the paper trail — receipts, exhibition records, certificates, expert attribution — that lets a buyer trust a piece is what it claims to be and worth what is being asked. Authenticity is the related question of whether the work is genuinely by the claimed author. Provenance is the evidence; authenticity is the conclusion you are able to draw from it.

Crucially, provenance has always been about trust under uncertainty. You usually cannot verify a work’s origin just by looking at it, so you rely on a documented chain instead. That logic, built for paintings and antiquities, turns out to be exactly what the digital creative economy now needs.

Why it traditionally mattered for value and trust

In any market where copies, fakes and misattributions are possible, provenance is what separates a valuable original from a convincing imitation. It supports price, enables resale, protects buyers from fraud, and gives the maker’s reputation something to attach to. Two visually identical objects can be worth wildly different amounts purely because one has a clean, documented chain to a known author and the other does not. The information about origin is itself the asset.

Why cheap, infinite imagery makes provenance more valuable

Generative AI has made high-quality imagery abundant and almost costless to produce. Abundance drives the price of the content itself toward zero — but it simultaneously raises the value of knowing where something truly came from. When anyone can generate something that looks like a hand-made original, the scarce and trustworthy thing is the verifiable link back to a real person and a real process. Counterintuitively, the cheaper synthetic content gets, the more a clean human provenance is worth.

There is a defensive dimension too. As convincing synthetic and manipulated media proliferate, buyers and platforms increasingly want a way to tell the difference, and trying to detect fakes after the fact is a losing game. The more durable approach is the inverse: proving authenticity at the source by attaching a tamper-evident record to genuine work, so the real thing can demonstrate its origin rather than every fake having to be caught.

Ways to establish authenticity

There is no single magic mechanism, but several reinforcing ones, and they work best in combination:

• Records and documentation: dated process files, drafts, working materials and a clear account of how a piece was made.
• Editions and certificates: numbered editions and certificates of authenticity that tie a specific item to a specific maker.
• A credible human body of work: a consistent, recognisable output over time is itself strong evidence of genuine authorship.
• Content provenance standards: emerging technical standards attach cryptographically signed metadata recording who created a file and what tools touched it, creating a tamper-evident chain of custody.
• Clean operational systems: keeping the line from finished artwork to finished product unbroken, so nothing in the pipeline muddies who actually made the art.

None of these is perfect on its own. Technical provenance standards, for instance, are powerful but only as good as their adoption, and metadata can be stripped if a file passes through tools that ignore it. A credible body of work is hard to fake but slow to build. The strength comes from layering: documentation, editions, reputation and technical records together make a claim that is genuinely hard to forge.

Why systems that preserve a clean chain matter now

All of this points to a practical conclusion. In a world of infinite generated imagery, the systems creators rely on should make it easier, not harder, to maintain a clean line back to a human author. Anything that blurs that line — by generating new imagery in a creator’s style, or by obscuring who actually made the art — quietly destroys the very provenance that is becoming the scarce, valuable thing. The architecture you build a creative business on now encodes a stance on authenticity, whether you intend it to or not.

Preserving that chain is the whole point of Realform’s design. Our ‘compose, never generate’ principle means the AI never produces imagery in a creator’s style; it takes the creator’s own finished, human-made artwork and composes it onto made-to-order products, then runs the business around it. Because the AI is pointed at the admin and never at the art, the provenance stays clean — there is always a clear, unbroken line from the product back to the human who actually made the work, and the copyright, credit and income stay with them.

Provenance used to be a specialist concern at the top of the art market. The age of generated imagery has democratised the problem and, with it, the value of solving it well. The creators who thrive will be those who can show, credibly and consistently, that there is a real person at the end of the chain — and who build on systems designed to keep that chain intact rather than quietly erase it.