Today a creator's only way to tell AI what it may do with their work is one switch in a text file — read at crawl time, by whoever bothers, and impossible to take back. Ask First moves that answer onto the creator's AI Passport and makes systems fetch it at the moment they act.
It is tempting to read this as an ethics failure — companies behaving badly, creators too small to stop them. That is not quite what is happening. Creators are not losing this because they said yes. They are losing it because there was never a way to be asked. Five mechanisms are missing, and every one of them makes taking cheaper than asking.
robots.txt has, by default, agreed to everything. The burden sits on the person with the
least power and the least warning — and it is a burden that has to be renewed on every platform they post to.Musicians, illustrators, essayists, voice actors, photographers, performers — anyone whose recognisable output is the asset. Crucially they are not anti-AI. They want indexing and citation, because that is distribution. They refuse voice cloning, because that is impersonation. Today they cannot say both.
Not the creators. Voice, likeness and generation platforms already cannot close enterprise deals without consent provenance — agencies demand indemnification. A vendor who can produce receipts can sell; one who cannot, increasingly cannot. Creators stay free, matching the Passport's free-for-life promise.
The primitive is general. Use-time resolution, expiring leases and content-free receipts are the same shape whether what travels is a licence, a workplace credential, an agent's spending limit, or an eligibility proof. We aimed it at creative rights because that is where the absence bites hardest today — and because a mechanism should be proven somewhere specific before it is sold as a platform.
A creator's honest position is nine independent decisions, and each one changes depending on who is asking. Almost all of them want to be findable — that is distribution. Almost none want their voice cloned — that is impersonation. There is currently no way to say “find me, but do not become me.” Here is the whole vocabulary they have today, next to the one they should have.
The published work carries only a pointer to the creator's Passport — never the permission itself. That single design choice is what makes withdrawal mean something: there is no stale copy of the answer anywhere to go out of date. These steps light up as the demo runs.
ego:junokade — inert, and useless on its own.Everything below is live. Change the policy on the left, resolve on the right, watch the answer change.
requester_class is asserted by Egoist against a credential, never read from the request body,
or an ad agency would simply declare itself a research lab.
// POST /creative/resolve // press Resolve to see the requester-facing response
The first question anyone serious will ask, answered before they have to ask it.
Consent protocols are either custody-based — the intermediary holds the asset and can withhold it — or declaration-based, holding nothing and instead making the permission state legible and non-repudiable.
This is declaration-based for public work, because the alternative means Egoist holding every creator's catalogue: a concentration risk worse than the problem. So a lease does not stop someone who already scraped a track. It changes what they can later claim — “the opt-out was ambiguous, we didn't know” stops working when there was a machine-readable answer at a known address.
It is custody-based where custody exists — voice models, likeness reference, unreleased stems. There the lease gates the actual bits. And nothing here un-trains a model. No consent layer can.
“Why would a scraper voluntarily call your resolver?” It wouldn't. You don't convince the scraper — you convince the scraper's customer.
Agencies and brands buying creative-AI tools already demand indemnification, and vendors already compete on licensed data. A vendor who can produce resolution receipts can sell into that market. One who can't, increasingly can't. The receipt isn't a compliance chore — it's the artifact they show the client.
Which answers who pays: requesters pay for consent provenance they can put in a contract. Creators pay nothing, matching the Passport's free-for-life promise.
The field is not empty and pretending otherwise would be the fastest way to lose credibility. IETF AI-preferences is standardising vocabulary past the single bit. RSL attaches licensing terms to content. C2PA carries data-mining assertions. Cloudflare pay-per-crawl puts a toll booth in the network path.
All four are anchored to content or origin and read once, at fetch. Creative Pass is anchored to a person and read at use — which is what makes withdrawal mean something and per-requester answers possible. It composes with all four; the pointer can ride inside a C2PA assertion.
First version: one voice or likeness platform with an enterprise sales problem, plus Egoist's
creator base onboarding at ego.ist/i/you. Not a standards coalition.