Présentation

Cette nouvelle session fait suite à l'édition très réussie de l'an dernier qui a rassemblé un grand nombre de professionnels venant de treize pays européens, du Brésil, du Canada, des Etats-Unis, de Chine et de Malaisie, ainsi que des hauts fonctionnaires d'institutions de renom.

L'intelligence artificielle (IA) et les robots font depuis longtemps l'objet de la science-fiction. Ce futur imaginaire est à présent devenu une réalité. Le marché de l'IA devrait passer de 8 milliards de dollars en 2016 à plus de 47 milliards de dollars en 2020. Les investissements dans l'IA ont augmenté de plus de 300% en 2017 par rapport à 2016. Des machines intelligentes, des algorithmes d'apprentissage automatique et des réseaux de neurones envahissent notre quotidien. L'interaction avec l'IA devient de plus en plus fréquente, en particulier dans l'environnement numérique. Ainsi, la réglementation des activités liées à l'intelligence artificielle est devenue un enjeu majeur de politique publique. Il devient dès lors impératif que les juristes, les entrepreneurs ou les décideurs maîtrisent les principaux défis juridiques de l'IA s'ils veulent jouer un rôle dans cette révolution qui occupera le débat pour les prochaines années à venir et au-delà.

En particulier, les effets perturbateurs de l'IA sur les modèles d'affaires traditionnels obligeront à reconsidérer le cadre de la propriété intellectuelle. Aujourd'hui, l'IA et les algorithmes écrivent des nouvelles et des romans, génèrent des œuvres d'art, écrivent et interprètent de la musique. L'intelligence automatisée peut stimuler les processus d'innovation de sa propre initiative. L'IA défie les notions juridiques les plus traditionnelles en matière de propriété intellectuelle, telles que la « copie », « l'originalité », le « créateur », « l'auteur » ou « l'inventivité ». Un robot peut-il être un auteur ? Un robot peut-il être un inventeur ? Un robot peut-il co-écrire un travail réalisé avec l'intelligence humaine ? Qui est titulaire des œuvres ou des inventions générées par l'IA ? Les inventions d'IA devraient-elles être considérées comme des antériorités ? A qui appartient l'ensemble de données à partir duquel une intelligence artificielle est réalisée ? Qui devrait être responsable de la créativité et de l'innovation générées par l'IA, si elles empiètent sur les droits d'autrui ou d'autres dispositions légales ?

Ce programme de formation explorera ces questions épineuses en analysant les aspects juridiques, politiques et éthiques concernant l'impact de l'IA sur la créativité et l'innovation. Dans ce cours, le CEIPI réunira des experts de premier plan dans le domaine pour discuter de ces questions nouvelles dans une perspective internationale plus large, en examinant diverses solutions envisagées au niveau national. Ce programme de formation devrait intéresser les économies créatives et les professionnels et entrepreneurs des hautes technologies, les juristes, les décideurs, les juges et les étudiants.

Programme

Day 1 — Thursday, 16 May 2019

9:00 : Welcome Address
Christophe Geiger, Professor of Law, Director General and Director of the Research Department, Center For International Intellectual Property Studies - CEIPI, University of Strasbourg, France
Giancarlo Frosio, Senior Lecturer and Researcher, Center For International Intellectual Property Studies - CEIPI, University of Strasbourg, France

9:30 : Module 1 - Overview - History and Theory of AI Regulation

Module 1 will provide an introduction to the training program in the first 30 minutes. The remainder of the class will discuss the theoretical, historical, and market landscape against which regulation of Artificial Intelligence (AI), Machine Learning (ML) and robots is emerging, with particular emphasis on machine-generated or computational creativity. AI's disruptive effects on traditional business models will force a re-consideration of the Intellectual Property (IP) framework. In this context, this module will first present legal tools available to protect AI, including trade secrets, copyright and patent protection. Later, it will briefly introduce the landscape of normative efforts in multiple jurisdictions that starts copying with a potentially ground-breaking revolution. This module will also consider how alternative approaches to computational creativity regulation will have far reaching commercial implications, shifting incentives for developing AI

Speaker : Giancarlo Frosio, CEIPI

10:45 : Coffee Break

11:15 : Module 2 - AI and Copyright - Authorship

Artificial intelligence writes poems, novels and news articles, composes music, edits photographs, creates video-games, and makes paintings and other artworks. AI can engage in any creative activities as technology like 3D printing enables computers to create physical artifacts without the need of human intervention. Like Google's Deep Mind, which generates and performs music or creates artworks, AI does so by listening to other music or analysing previous artworks online. Which are the conditions for protection of creations generated by deep neural networks under the main copyright regimes? Is AI an author according to tradition copyright standards ? Should traditional copyright standards such as originality apply, and perhaps machine-generated creative works fall in the public domain? This module will try to provide an answer to these basic issues surrounding AI's creativity by looking into legislation implemented in several jurisdictions and relevant case law

Speaker : Ana Ramalho, Assistant Professor, Maastricht University

13:00 : Lunch

14:00 : Module 3 - AI and Copyright : Ownership and Infringement

After reviewing standards for AI's authorship, Module 3 will delve into complex matters related to ownership of machine-created works and infringement. Who owns the copyright in a work generated by a machine ? Should specific arrangements conferring authorship to the agents spending skills, labour and efforts to create AI in the first place regulate the field ? In this context, however, ownership might still be tricky to allocate. Does it belong to the person who built the system, the person who trained it, or the person who fed it specific inputs ? Again, AI might engage into copyright infringement as a result of its creative activities. How does the dichotomy idea/expression, the notion of originality or the doctrine of fair use apply to computational creativity ? Open questions become more complex in light of the growing power of ML algorithms to rewrite reality. ML tools can turn shots of horses into zebras, black bears into pandas, dogs into cats, apples into oranges, and porn stars into celebrities, multiplying grounds for violation of economic and moral authorship rights and personality rights. Where to cast relevant liability for infringement in all these cases ?

Speaker : Ana Ramalho, Assistant Professor, Maastricht University

15:30 : Coffee Break

16:00 : Module 4 - AI aAnd Copyright : Automated Copyright Enforcement

This module will focus on AI's applications in content moderation on digital platforms. In particular, copyright enforcement has been increasingly dealt through automated filtering and other algorithmic means. While semiotic governance online has become an issue that calls for extreme measures, taking down content through automated means poses challenges for online expression and access to information. In this scenario, governments and policy-makers are heavily pressuring companies to take action and a few jurisdictions have already responded with new regulatory initiatives. Meanwhile, judicial decisions have highlighted the inconsistences between automated enforcement and fundamental rights

Speaker : Giancarlo Frosio, CEIPI

17:30 : Welcome Cocktail

Day 2 — Friday, 17 May 2019

9:00 : Module 5 - AI, Data and Big Data - Ownership and Protection

Module 5 will look into novel issues emerging in connection with AI and data management. Data and Big Data processing is indeed a fundamental portion of machine learning. On one side, data ownership might emerge as critical issue to be carefully considered when dealing with AI and ML systems. Developing AI and ML systems generally involves training it using large datasets, so the system can continuously improve its decision-making abilities. Who owns the IP in the datasets which are used to train the system ? Although data might be freely available online, it cannot be used for any purpose. Therefore, genuine issues of liability for use of proprietary data in ML processes might arise. On the other side, data protection regulations will play an important role in the evolution of AI and ML systems. This section will consider relevant legislation and case law, with particular emphasis on the interpretation of the upcoming EU General Data Protection Regulation's provisions on profiling and automated decision-making

Speaker : Andres Guadamuz, Senior Lecturer, University of Sussex, United Kingdom

10:30 : Coffee Break

11:00 : Module 6 - Patenting AI

Module 6 will consider a vast array of issues related with patenting AI and ML systems. In this context, this module will review international legislation, case law and patent offices' practices, with special emphasis on EU-US comparative analysis. First, a fundamental challenge for protecting AI technologies with patents involves claiming subject matter that is patent eligible. Also, this module will consider how to identify what contributed to the development of an AI-related patent for the purposes of determining whether someone was an inventor. Further, satisfying disclosure requirements can be challenging when seeking patent protection for AI-based inventions. What should be disclosed in AI inventions to meet the requirements ? Again, how an AI-based invention claim should be drafted ? How does the doctrine of equivalents apply to AI inventions ? Do different standards apply to rule-based systems and neural networks ?

Speaker : Jean-Marc Deltorn, Patent examiner, European Patent Office, EPO

12:30 : Lunch

14:00 : Module 7 - AI Generated Inventions - Inventiveness and Ownership

What if an AI-enabled machine invents something ? What if an AI algorithm—without any human intervention—develops a new business method, a drug, a machine, or other invention ? What if an AI develops a technical improvement of itself ? In this respect, as well as in the copyright domain, AI challenges the most basic patent notions. Can a robot be an inventor ? Who owns AI generated inventions ? This section will present the conditions under which the products of AI processes/systems can be granted protection. The issue of industrial applicability and plausibility will be discussed, as well as the relation to article 64(2) EPC (when the AI parent process is patentable). The problem of inventive step will be discussed too as well as the issue of defining the person skilled in the art when machines are imbued with an increased level of autonomy and (technical) creativity. The second part of the presentation will be devoted to the issue of inventorship and the (necessary) mention of the inventor on the title. This section will also expand upon the identification of the inventors and the difference between US and EU

AI Generated Inventions - Non Obviousness and Prior Art

As part of the “All Prior Art” project, an AI system continuously produces any possible meaningful combinations from published patents. A sister website “All The Claims” is attempting the same thing, but with the use of claims. These projects presumably aim at making the published concepts not patentable. Algorithms, AI and deep learning could potentially make everything under the sun prior art. This might support the idea that AI generated inventions without human intervention should be generally excluded from prior art. Is this the case ? To which extent publication of computer-generated content should be treated as prior art and allowed to prevent others from obtaining patent protection on independently created inventions ? Under current law and practices, do automatically generated claims qualify as prior art ? Should automatically generated claims qualify as prior art ?

Speaker : Jean-Marc Deltorn, EPO

15:30 : Coffee Break

16:00 : Module 8 - AI and Patent Enforcement

In the same way that AI challenges traditional concepts of authorship and inventorship, it also raises fundamental challenges to the concept of enforcement. The latter is generally understood as requiring the participation of the right holder and public authorities and is characterised by the implementation of fair trial basic tests. This session explores the possibilities arising from devices that automatically enforce intellectual property rights and what does self-enforcement mean for legal theory and in daily practice. This session will also present advances in deep learning relating to IP enforcement and how practice in courts and legal firms is being transformed by data analytics and AI

Speaker : Xavier Seuba, Associate Professor, CEIPI

Day 3 — Satuday, 18 May 2019

9:00 : Module 9 - AI, Trade Secrets, and Medical Innovation

This module will consider trade secrets as an additional legal tool for protecting AI. This module will contrast patent protection for AI inventions with trade secrets protection and consider potential shortcomings of patent protection, in general. Unlike a patent, whose granting period might take a few years, trade secret protection arises automatically if secrecy of information creates a competitive advantage and there are reasonable measures in place to maintain secrecy. In this regard, trade secret protection may be especially well-suited for fast developing and changing AI inventions, whose improvements occurs at an extremely rapid pace. This module will consider the international protection of trade secrecy for AI by looking into recently adopted EU trade secrets directive, with some legislative examples. In particular, this module will discuss a case study where a trade secret and patent protection plays an increasingly interchanging role in relation to medical innovation, with special emphasis on personalized medicine

Speaker : Nari Lee, Professor, Hanken School of Economics, Finland

10:30 : Coffee Break

11:00 : Module 10 - Autonomous Driving

This module will discuss connected and autonomous vehicles as an emerging field where AI might raise relevant IP, trade secrets, and liability issues. This module will first introduce the landscape of autonomous driving projects and market applications. It will map out present legislation and emerging policy proposals to regulate autonomous driving. This module will discuss relevant issues related to the licensing, transfer and protection of autonomous driving technology. IP rights will play a critical role in enabling industry players to establish, and maintain, a position within this emerging market, with thousands of possibly competing patents being sought. The race to achieve market share will inevitably lead to a flurry of IP disputes, on the basis of patent, trademark, design, copyright or trade secrets infringements, as the Uber/WAYMO case might already show. Finally, this module will consider in details, IP liability that might arise from autonomous driving enabled machines. Under which standards liability should arise? To whom liability should be attributed for damages caused by machines and vehicles autonomously operated by an AI? These are all very novel questions that have been considered so far only from a theoretical perspective but shall have soon to be tested in practice

Speaker : Nari Lee, Hanken School of Economics

12:30 : Lunch

14:00 : Module 11 - Blockchain

Module 11 will focus on blockchain technology and alternative, decentralized architectures that rely on peer-to-peer networks and distributed technologies to provide secure and autonomous platforms for online interactions and communications (BitTorent, Bitcoin, Ethereum, etc). This module will analyse the legal framework in which these platforms operate, as well as alternative governance models combining regulation by code, contracts and social norms. « Distributed ledger technologies » (of which the blockchain is one instantiation) will also be discussed in relation to patentability (and exclusion thereof) either as mathematical method, business methods, or computer programs. In addition, this module will highlight how blockchain technology provides opportunities for both infringement and enforcement. Blockchain allows to track ownership/transactions, effect payments, integrate data, and provide transparency. Existing implementation of blockchain technology for IP management include inter alia the music platform Muse and Ujo Music or the blockchain patent exchange Kyna. Finally, this module will consider liability that might emerge from managing and using these technologies, especially in the context of financial transactions

Speaker : Jean-Marc Deltorn, EPO
Speaker : Andres Guadamuz, Sussex

15:30 : Farewell

Pour toute question veuillez contacter Mme Isabelle Mahoudeau pour des questions d'ordre administratif/inscriptions : imahoudeau@ceipi.edu
M. Giancarlo Frosio pour des questions relatives au contenu pédagogique du programme : gcfrosio@ceipi.edu