On 11 May, the European Parliament’s joint lead committees adopted a compromise version of the bloc’s Artificial Intelligence Act (AI Act) by a substantial majority.
The updated version of the bill bans the use of real-time and post hoc facial recognition in public spaces, as well as AI systems that deploy subliminal techniques designed to be purposefully harmful through manipulation or deceit.
While the text includes a general carve-out for free and open-source AI, the latter will only fall under its jurisdiction if it is used in the context of high-risk or generative AI (foundational models). Specific obligations for this are outlined in the text.
Following a final Parliamentary vote that is currently expected during the week of 12 June, trilogue negotiations between the European Parliament, the European Commission, and the Council will begin.
At the end of June, Spain will take over from Sweden as Council Presidency. Spain strongly intends to reach a final agreement between all three institutions by the end of 2023. However, the upcoming trilogues are likely to witness contentious debates on issues including open source, generative AI, and facial recognition.
If you want to learn about how the EU plans to regulate generative AI and its impact on the creative industry, you can attend our webinar “Generative AI & the Creative Sector: The EU’s AI Act” on May 23, 2023 at 14:00 – 15:00 BST.
Access Partnership is closely monitoring regulatory developments related to the EU AI Act. If you want more information, you can contact Lydia Dettling at [email protected].