Workshop: Legal aspects of conversational AI
Time:
Monday 9 September 2024
13:30 - 15:30
Location:
Ústav státu a práva AV, Národní 18, Praha 1, 7th floor
The workshop will be held in English and in person only.
The workshop will take place in the project supported by The Technology Agency of the Czech Republic No. TQ12000040.
Programme:
Extra-contractual Liability for Large Language Models in the EU
Prof. Dr. Martin Ebers
The advent of Large Language Models (LLMs), such as ChatGPT, raises the question of whether the EU legal framework adequately addresses extra-contractual liability for damage caused by these systems. The presentation maps the distinctive features of LLMs as compared to other digital technologies as well as to "traditional" AI systems. Against this background, the paper examines various existing and forthcoming liability regimes - such as the proposals for an AI Liability Directive and a revised Product Liability Directive, the AI Act, non-discrimination law and the GDPR - to assess their sufficiency and effectiveness in the light of LLMs.
Navigating Privacy Issues in gathering Training Data for AI: a Case Study of Google
Martin Erlebach
At the beginning of July 2023, Google changed a part of its privacy policy to read: “For example, we use publicly available information to help train Google’s AI models and build products and features like Google Translate, Bard, and Cloud AI capabilities.” – effectively stating that it will scrape everything that is posted online for its benefit. In the current environment of many generative AIs appearing almost daily, the lawful acquisition of training data deserves close attention. This recent change in Google's privacy policy may offer significant insight into the legality of keeping up with AI’s data-crunching habits. This presentation will explore current ways of gathering training data for AI as well as privacy concerns when doing so. To illustrate the ways training data is currently gathered a case study of Google scraping the internet for training data will be presented. The case study concludes that the current practice of one of the largest players in the market is in breach of GDPR and how this may reflect on the practices of other companies as well as reactions of the public to this fact.
AI and Copyright
Dr. Eva Fialová
Artificial intelligence systems generate outputs of different nature. The "creation" of artificial intelligence raises a number of questions. The problem of authorship of AI output is the most frequently mentioned. That is, whether the author can be the AI itself, or whether it is the prompt (query or assignment) or someone else. Before the question of authorship itself, it is necessary to ask whether the output of an AI is an author‘s work in the sense of copyright law, i.e. whether the output is a unique result of creative activity. Artificial intelligence needs data to train and improve itself. Data can and often does have the character of a copyrighted work. In order to use a copyright work, the provider of an AI system would need to have a licence. It can be assumed that most copyright works have been used to train AI without a license. While there is a legal license in copyright law for text and data mining, there are limits to its use for training AI. This paper discusses the problems that the use of AI brings to copyright law, in particular the issue of authorship and the legal regime of training data.
Prof. Dr. Martin Ebers is President of the Robotics & AI Law Society (RAILS), Germany, and Professor of IT Law at the University of Tartu, Estonia. Moreover, he is permanent fellow at the law faculty of the Humboldt University of Berlin, and co-director of the German Institute for Energy and Competition Law in the Public Sector. In 2022, Dr. Ebers was awarded a five-year grant from the Wallenberg Foundation (WASP-HS) to conduct research with his team in Sweden on “Private Rule-making and European Governance of AI and Robotics”.
His latest books are amongst others Algorithms and Law (Cambridge University Press, 2020); Rechtshandbuch Künstliche Intelligenz und Robotik (C.H. Beck Publishing, 2020); Algorithmic Governance and Governance of Algorithms (Springer Nature, 2021); Contracting and Contract Law in the Age of Artificial Intelligence (Hart Publishing, 2022); and Stichwortkommentar Legal Tech (Nomos Publishing, 2023).
Martin Erlebach is a PhD student at the Institute of Law and Technology Masaryk University. His current research focuses is data monopolies, AI and personal data protection.
Dr. Eva Fialová is a researcher at the Institute of State and Law of the Academy of Sciences of the Czech Republic. In the past, she collaborated with the Institute of Law and Technology of the Faculty of Law of the Masaryk University. Her field of research is information technology law, especially in the area of algorithms and autonomous systems application and data protection. Eva Fialová is the author of a book “Contactless Chips and Privacy Protection,“ co-author of the books "Civil Liability for Operation of Autonomous Vehicles", “Profiling and automated decision making (not only) in the light of human rights and fundamental freedoms“ and “Protection of personality, privacy and personal data“, well as the author of scientific articles concerning various aspects of modern technologies and the law. Eva Fialová obtained her Ph.D. at the Masaryk University in Brno. Before that she studied master programme at the University of Amsterdam.
Registrační formulář