There’s more and more AI — but far too little control.


AI has arrived in everyday administrative life. According to the OECD, it’s now used in at least one area of government in 35 of 36 OECD states. Great! But there’s a catch: only 14 of 36 countries even require a mandatory risk analysis before using AI. Only 11 of 36 have formal transparency standards for algorithms. And only 6 of 36 keep an open algorithm register.

That means: there’s far too little control. And there are far too few ways for businesses or citizens to get insight into how AI is being used. That’s explosive, because AI isn’t a toy — in doubtful cases, it decides on things that can change people’s lives. And if you don’t get that under control, you get what happened in the Netherlands, where thousands of households received enormous repayment demands — because an AI had made a mistake.

So it’s not enough to simply “introduce” AI. It has to be made transparent, understandable, and always kept under human oversight too.


Image: This image was created with AI.

Source:
OECD, Digital Government Outlook 2026, chapter “Adopting and governing AI in government”

You can find this post on LinkedIn here: https://www.linkedin.com/posts/janwegener_digitalisierung-verwaltung-ki-activity-7477986905721122816-vXCK?utm_source=share&utm_medium=member_desktop&rcm=ACoAAF-yDsQBgbICDl3wlRGoCQSZHiHxHemNJR8