AI Can Now Write Laws — But Nobody Knows Who's Responsible

Two legal scholars, Julieta Rabanos and Bojan Spaić, have published a new paper asking a question most people haven't considered: can a machine actually write law? Not just legal-sounding text, but something that counts as a real rule with real consequences. Their answer, roughly, is yes — and that's a bigger deal than it sounds.

Here's the problem they focus on. Law has always assumed an author: a human lawmaker, judge, or official who intended to create a rule and meant something specific by it. When courts argue over a confusing clause, they ask what the writer was trying to say. ChatGPT-style AI has no intentions at all. It predicts likely words. It doesn't want anything, mean anything, or understand anything. So how can its output be law?

The surprising part is where the authors land. They argue that intention was never as essential as lawyers like to think. People can read a rule and make sense of it even if nobody 'meant' it. In that view, AI can generate legal text, and humans can interpret it into binding rules. The author may be dead — but the law lives on.

For you, this isn't abstract. AI is already drafting contracts, terms of service, insurance policies, and court filings, and cities have experimented with AI-written ordinances. Millions of people click 'agree' on AI-assisted documents every day. The paper's real warning is about accountability: if a harmful clause was never intended by anyone, who do you sue, appeal to, or vote out of office? Regulators and judges will have to answer that soon.