Anyone who has ever tried to open a legal textbook—or even just read an official document drafted by any administrative body—knows how much the language of the law resembles a dead tongue that, mockingly, continues to legislate over the living. 

It’s as if legal norms insist on speaking in cryptic hendecasyllables, hiding the vital information they aim to communicate within the folds of convoluted subordinate clauses.

Enter Aptus.AI, a startup based in Pisa, which has launched its Manifesto for AI-Proof Regulation with a proposal that appears at once simple and revolutionary: make laws machine-readable. This is not just a legal nerd’s whim, but a way to transform the law into a more transparent, traceable, intelligible ecosystem—not just for humans.

Behind the technicality of the “machine-readable format” lies a deeper ambition: to train artificial intelligence to understand, explain, and even help write legal norms. This is made possible by Akoma Ntoso, an XML standard developed in Bologna (named in Akan with a poetic expression: “linked hearts”), already adopted by international institutions and, to some extent, by the Italian Senate. The rest of the country’s legal production, however, remains trapped in outdated formats—barely digestible even for a large language model, let alone an average citizen.