Digdeep Podcast: Well-Intentioned but Dangerous – Will AI Regulation Become a Tool for Censorship?

In this episode of #digdeep, MCML Junior Member Sarah Ball talks together with Phil Hackemann about AI regulation.

Most people want AI to be safe, but the same mechanisms designed to protect AI from misinformation and harmful content can also be used for censorship. As AI increasingly becomes a central gateway to information, those who decide what models are allowed to say gain significant influence over society’s information space.

In their award-winning paper, Sarah Ball and Phil Hackemann show that this power is currently concentrated among a small number of AI providers and governments. They explain where AI models can be controlled, why Elon Musk’s Grok serves as an example of manipulation, and why open-weight models alone do not guarantee digital sovereignty.

Perhaps most intriguing is the question of what happens when AI models begin to reflect on their own control and behavior.

Just listen in (podcast in German).