Ownership turns on human creative control.
Not on how much you typed, or which tool you opened. This pillar covers what is protectable, what has to be disclosed, and who is accountable when it goes wrong.
Everything in this pillar
Ownership, disclosure wording, publisher policies, and accountability when AI-assisted text is wrong.
The first guides in this pillar are being written now.
Four things worth getting right
Drawn from copyright guidance and from published work on how journals and publishers actually govern AI use.
Creative control is the test
Copyright guidance asks how far a human controlled the work's expression and actually formed the traditional elements of authorship — the same test applied to a photographer using Photoshop.
Disclosure is becoming mandatory
Responsible-use frameworks call for required disclosure plus documentation of provenance — not a vague acknowledgement at the end.
Authorship is still contested
Whether a model can be listed as an author remains an open argument in publishing, and one where policies differ sharply between venues.
Policy belongs in the workflow
Governance embedded only in author guidelines misses where AI actually enters — drafting, review, editing and production alike.
Questions people actually ask
Can I copyright something I wrote with AI?
The question is not whether AI touched it, but how much of the expression you actually formed. Copyright guidance draws the line at human creative control — a work where you selected, arranged, edited and shaped the expression can be protectable, while raw model output prompted into existence generally is not.
Do I have to say I used AI?
Increasingly yes, and the direction of travel is toward mandatory disclosure with documented provenance rather than a vague note. What varies is the venue: journals, employers and platforms each set their own threshold, so the practical answer is to check the specific policy that binds you.
Who is responsible if AI-assisted text turns out to be wrong?
You are. Every governance framework that addresses this puts accountability on the human author, which is precisely why disclosure and verification are treated as paired obligations rather than alternatives.