“So we cannot know whether artificial intelligence will help us infinitely, whether it will ignore us and push us aside, or whether it will possibly destroy us,” famous physicist Stephen Hawking said back in 2017. The latest hype about the use of artificial intelligence as chatbots in search engines, namely ChatGPT in Microsoft’s Bing or Bard on Google, highlights the current developments in the field of AI. Although the legal consideration of these phenomena is still in its infancy, initial discussions are already being vehemently conducted. Chatbots are an illustrative example of using AI to approach some key AI legal issues.
At EU level, work is currently underway on the “Artificial Intelligence Act”, or AI Act for short. The regulation is about to be finalized by the EU Commission, EU Council and EU Parliament. Like the General Data Protection Regulation, it will apply immediately in all EU countries after it has come into effect, without the need for national implementation. It is expected to come into force in 2023, as the EU institutions will not have much time to implement their agenda due to the upcoming European elections in 2024. If this does not succeed, there is a risk of a considerable delay in the uniform regulation of AI in the EU.
Tobias Haar is a lawyer specializing in IT law at Vogel & Partner in Karlsruhe. He also studied legal informatics and holds an MBA.
The EU describes the area of tension in which the AI Act is intended to regulate as follows: “The EU’s approach to artificial intelligence focuses on excellence and trust in order to strengthen research and industrial capacities and to guarantee fundamental rights.” The text of the ordinance has been fine-tuned for two years. Over 3000 amendments were tabled. Even the definition of AI is controversial. A current proposal is to limit the scope to systems “developed through machine learning techniques and knowledge-based approaches”.
