Machines Learning the Rule of Law EU Proposes the World’s first Artificial Intelligence Act

On 21 April 2021, the European Commission (EC) proposed the world’s first Artificial Intelligence Act (AIA). The proposal has received a warm welcome across the EU as well as from the US, as it includes substantial legal provisions on ethical standards. After its release, the media’s main focus laid on the proposal’s “Brussels Effect”, which refers to the EU’s global regulatory influence: EU laws exceed their “local” influence and become global standards. With the AIA, the EU has the potential to become the world’s “super-regulator” on AI.

More than the Brussels Effect, however, the emphasis should lie on the EU’s intention to explicitly protect the rule of law against the “rule of technology”. Despite this expressed goal, the normative power of the regulation to ensure the protection of the rule of law seems inadequate and raises serious concerns from the perspective of fundamental rights protection. This shortcoming becomes most evident across three main aspects of the AIA, namely in the regulation’s definition of AI systems, the AI practices it prohibits, and the preeminence of a risk-based approach