Issue Brief: The EU Artificial Intelligence Act
INTRODUCTION
European Union (EU) legislators are considering legislation— the Artificial Intelligence Act (AIA), the original draft of which was published by the European Commission in April 2021[1]—that aims to ensure the safety of AI systems in uses designated as “high risk”. As originally drafted, however, the AIA’s scope was not at all limited to AI; it would instead cover virtually all software. EU governments seem to have realized this problem and are trying to fix the proposal, while some pressure groups have pushed to move the draft in the opposite direction.
The AIA proposal is currently under consideration by specialized committees of the European Parliament. The parliamentary stage began with a long disagreement among the various committees regarding who should have decisive influence over the Parliament’s position on the bill. With that disagreement now resolved, discussions on the legislation’s merits are ongoing.
The purpose of this brief is to inform debate on the proposal’s fundamental features: its scope and the key provisions setting out prohibited AI practices (related to so-called “subliminal techniques” and “social scoring”).
Read the full issue brief here.
[1] Proposal for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) and Amending Certain Union Legislative Acts, European Commission, (Apr. 21, 2021), available at https://perma.cc/RWT9-9D97.