Consensus without a champion: Cross-partisan demand for AI safety regulation in Hungary and the missing belief in delivery

Industry leaders now ask for independent, pre-release safety testing of frontier AI systems; the US president dismissed such calls as fears about things that “won’t happen.” What the public wants is well documented for the US and UK, but almost unknown elsewhere, and little is known about whether people believe their preference will be acted upon. In a Hungarian online sample (n = 1,356), 75% supported mandatory independent safety testing before AI systems are released, regardless of partisanship. Three in four also believed most Hungarians agree, and 61% believed that this majority was growing. Yet only 32% expected such testing to be introduced. In one of Europe’s most polarized countries, AI safety regulation thus has a consensus but no champion. The missing link is not a hidden majority but the belief that anyone will deliver it.