EU reaches key agreement on AI regulation
On Friday 8 December, negotiators from the European Parliament and the Council reached a provisional agreement on the Artificial Intelligence (AI) Act, which sets out a legal framework to ensure the safe and fair use of AI.
On Friday 8 December, negotiators from the European Parliament and the Council reached a provisional agreement on the Artificial Intelligence Act, which sets out a legal framework to ensure the safe and fair use of AI. The harmonised text of the regulation prohibits high-risk AI practices such as social scoring and non-discriminatory face recognition, while allowing for the limited use of biometric recognition by law enforcement authorities under strict supervision.
High-risk AI systems, including those that could influence electoral outcomes, will require a mandatory fundamental rights impact assessment. The Regulation will also provide mechanisms for citizens to lodge complaints and request clarifications on decisions taken by and affecting high-risk AI systems.
To foster innovation, the text of the regulation supports regulatory sandboxes for AI development, with high penalties for non-compliance. This legislation represents a milestone for the EU in regulating the ethical use of AI, and the text will now progress to the formal adoption stages.
ODIPI is organizing ERA KR21 Conference: Barriers and Incentives for Open Science in the Copyright Law that will take place on 2 December, 2024 at Hotel Four Points by Sheraton (Mons) in Ljubljana and also online.
The District Court of Hamburg ruled in the case of Kneschke v. LAION e.V. that LAION did not infringe the copyright of photographer Kneschke, as the use of his photograph was covered by the exception for text and data mining (TDM) for scientific purposes.
“Can copyright bring artificial intelligence to its knees? Which other circumstances may cause that the “making” of generative AI can dramatically change in the (near) future. This short paper presents potential challenges that copyright poses to the training of the machines on large amount of data. Different jurisdictions address these issues differently. In the USA the legality of these activities is tested in several court cases. Do gentlemen’s agreements and pragmatic symbiosis known from the “search engines business model” provide sufficient basis and/or incentive for the business model of “making” generative AI business model as well?