Bavaria rejects an AI pause and calls for risk-based rules

Bavaria should pursue artificial intelligence while setting safeguards against its risks, Digital Minister Fabian Mehring said at a summit in Munich, BR reports. He rejected a pause in AI development in Europe, arguing it would not make the technology safer.

Mehring said stopping development only in Europe could leave the region dependent on rules set by the United States or China. He called for greater digital independence and “risk-based standards” instead of what he described as overly bureaucratic European regulation.

AI has drawn attention both for its industrial potential and for safety concerns. The French AI company Mistral recently opened a site in Munich, where research teams are expected to focus on industrial applications. Mistral named BMW as a partner, including on AI-simulated crash scenarios.

Meanwhile, OpenAI recently withdrew the release of a new model over safety concerns. A recently published report also described AI agents operating without human intervention and hacking the servers of another AI platform. Developers and managers have warned of the risks, with some calling for development to stop.

Mehring said Bavaria would set up an AI safety council and an AI ethics forum this year. He called for dialogue among technology companies, researchers and representatives of social groups, and said public debate was essential to ensure people were not afraid of digital change.

He also pointed to Pope Leo XIV’s May encyclical, Magnifica Humanitas, which addresses human dignity in the digital age. Mehring said its discussion of managing technology’s risks was not technophobic and expressed a wish for public debate to be conducted in a similar way.

Mehring said Bavaria would seek influence through Europe, bringing the results of discussions in the state to talks among digital ministers in Brussels. The Greens in the Bavarian state parliament said his proposals did not go far enough. Digital affairs spokesperson Benjamin Adjei said it was not enough merely to help shape safeguards.