This being Dutch politicians dealing with technology it wouldn't surprise me if they completely miss the point. In fact by basically only mentioning ChatGPT the

This being Dutch politicians dealing with technology it wouldn't surprise me if they completely miss the point. In fact by basically only mentioning ChatGPT they may have already. You can do quite a lot of harm with very simple models, all you need is unclear rules, little oversight and no avenues for escalation. Which in fact did go wrong, terribly, plunging tens of thousands of households into bankruptcy. You don't need something capable of modelling human language to make things worse. At most a more powerful model can simply do more, but the risks are roughly the same. And if neural networks writing obtuse language with little empathy following a complex set of rules are causing problems then I've got some bad news about Dutch officials.