Story here. It was in Canada. I suppose it was inevitable given how many others have relied on AI-hallucinated citations without checking. I suspect it has happened in the US as well, though perhaps without anyone knowing. Of course, if arbitrators don’t write opinions but merely ask AI questions and don’t verify the cites, no one would know the decisions could be challenged for being based on non-existent law. Which is both a reason to require arbitrators to produce written opinions and an incentive for them not to do so unless such a requirement is imposed.
