Ask a room of founders how copyright works and most will describe a registry. You make something, you file it somewhere official, you get a number back, and that number is what you point at if somebody copies you.
That is roughly how it works in the United States. It is not how it works almost anywhere else.
The rule almost nobody states plainly
Under the Berne Convention, copyright protection arises automatically the moment a work is created and fixed in some form. No registration, no filing, no fee, no formality of any kind. Berne has over 180 member countries, which covers essentially every jurisdiction with a functioning legal system.
This is genuinely good news and it is why almost nobody thinks about it. You own the copyright in the draft you wrote this morning. You did not have to do anything to get it, and nobody can require you to.
So why does the United States run a register
Because US law attaches specific procedural consequences to registration, not because registration creates the copyright. The copyright exists either way.
What registration does in the United States is act as a precondition for certain remedies and for bringing an infringement action over a US work. Register early enough and additional categories of damages come into play. Register late, or not at all, and you still own the work, but your options in a US court narrow.
That is a procedural rule of one national system. It has produced a widespread assumption that a copyright register is how copyright works everywhere, which it is not.
What automatic protection does not give you
Berne gives you the right. It does not give you the receipt.
Nothing in the automatic-protection rule creates a record of what you made, when you made it, or that it was you. The right attaches silently, at the moment of creation, with no witness and no artefact. In the ordinary case that is fine, because nobody disputes it.
The problem shows up in the case that is not ordinary.
What happens in a real dispute
The question in a copyright dispute is almost never "do you have copyright in this". Everyone accepts that whoever made it owns it. The question is who made it, and who made it first.
Two agencies pitch a campaign concept to the same client. A contractor reuses a deliverable for a competitor. A former employee launches with a product design that looks familiar. In each case both sides claim authorship, both sides are sincere, and the outcome turns on evidence of sequence rather than on the law of ownership.
At that point automatic protection has done everything it can do for you, and the burden moves to what you can show.
What people use instead, and where each one gives way
Posting it to yourself. The sealed envelope with a postmark. It is cheap and the instinct behind it is right: get a dated, third-party mark on the work. The weakness is that an unopened envelope proves an envelope was posted on a date, not what was inside it, and the method is old enough that its failure modes are well known.
Blockchain hashes. Better in one respect, because a hash genuinely binds to the exact file. The gap is the status of the timestamp rather than the maths. A public ledger entry shows a hash existed at a block height. It is not a qualified electronic timestamp issued by a Qualified Trust Service Provider, and it does not carry the legal presumption that one does. We set the two side by side in the comparison with blockchain timestamping.
A notary. Strong, and the right answer for a single high-value work. It attests that a document was presented on a date. The limits are practical: per-document cost, appointments, and it captures a print-out rather than binding cryptographically to a digital file. It does not scale to what a working studio produces in a week.
Platform upload dates. The most common answer and the weakest. The date is held by a company you do not control, under terms that change, in an account that can be closed, and it records when you uploaded rather than when you created. Useful as a signal. Not evidence you would want to rely on alone.
What a register outside the United States would have to do
The gap is real, but a second national register would not close it. Copyright is territorial, and a record that only means something in the country that issued it inherits the same limitation the US register has.
Something useful would need to be portable rather than national, so the proof travels with the file into any Berne country. It would need to bind to the exact work by hash rather than to a title or a description, since disputes turn on which version. It would need a time source somebody else vouches for, because a date you control is a date you can be asked to justify. And it would need to be checkable by a third party without contacting you, since the person who needs convincing is precisely the person who has decided not to take your word for it.
None of that requires a government to build anything. It requires the record to be made at the moment of creation rather than reconstructed afterwards, which is the part that is genuinely hard, because at the moment of creation nobody believes they will ever need it.
That is the whole problem in one sentence. The evidence is cheap to create on the day and expensive to assemble a year later, and you find out which you have on the day somebody disputes it. The mechanics of producing that record are set out on our Article 50 page, which covers the same evidence question from the AI-disclosure side.
