Intellectual Property Law Weak Artificial Intelligence Software Regulation

The basis of modern AI systems is the development of the mid-20 th century: algorithms and mathematical methods.The work does not consider strong Artificial Intelligence and its theoretical legal regulation.It is proposed to study the issue of legal regulation of weak Artificial Intelligence as one of the existing varieties of software,in particular artificial neural networks.Accordingly,artificial intelligence in this work implies weak artificial intelligence as a generalizing concept for software implementations of algorithms based on deep learning,as well as their hardware and software implementations.Since deep learning methods are used in completely different areas of development(including management,medicine,business,etc.),the task of the study does not include a detailed study of the legal protection of the use of AI in each separate sphere.This paper analyzed how to apply modern principles of law to the protection of innovative technological products in the field of AI-software and AI-related inventions.The goal is to understand which legal regimes of intellectual property provide the most balanced protection of AI systems,as well as understand the trends of legal development of intellectual property in the context of rapid technological progress.The angle is on comparing the Russian system with other Intellectual Property regulations,especially EU model,and with the legal doctrine and taking into consideration the technical aspects of issue.Legal protection of intellectual property in the IT industry is an ambiguous and complicated issue.The rapid development and complexity of IT technologies and the conservatism of legal regimes create gaps in legal regulation,and unbalanced legal regulation can lead to economic distortion and slowdown of technological progress.Strengthening the patent regime in the field of IT can result in weakening competition,while copyright does not always provide sufficient protection for the software code and its related objects.With the growth of automation in the field traditionally protected by copyright,it is increasingly difficult to rely on such categories of copyright as creativity,originality and authorship.Weak artificial intelligence is a technology that simulates one or another aspect of human activity,consisting of sets of algorithms executed by a particular programming language(as well as object code),and functioning on the basis of hardware.The ability to automatically generate products of intellectual activity such as databases,texts,music,images,etc.even more show the ambiguous position of copyright in the face of new technologies.Automation of the creation of databases leads to the uncontrolled processes in their regulation and protection.Growing poorly controlled data flows were the reason for the restrictive policy on the protection of personal data in the EU(GDPR).For data manufacturers and consumers,it means that the data has increased in value,and therefore the mechanisms for obtaining this data,based on deep learning algorithms,need both technical and better legal protection.The economic reasonability of protecting data obtained in a "non-creative" way gives rise to the reasonability of strong legal protection.From a technical point of view,artificial intelligence is software code,usually trained on a database,capable of simulating certain aspects of human activity.Modern AI tool systems are based on artificial neural network technologies and deep learning methods.Accordingly,the objects of legal protection are similar to the objects of regular software,although they have their own specifics.In particular,for AI created on the basis of deep learning technology,it is necessary to distinguish a trained network and an untrained network.In other words,the development of such AI does not end at the stage of writing the source code and its subsequent compilation.Neural network training requires a large database on which the neural network will be trained.Accordingly,the database for the neural network is an integral part of its structure.On the other hand,some neural networks generate databases by collecting data automatically.Thus,a database can be both a derived and a producing element of a neural network.Therefore,two different objects of legal protection should be noted: a database protected by copyright and a database protected by sui generis law.The functional-expressive dichotomy of the program code corresponds to the dichotomy of copyright and patent law.The dilemma of protecting ideas and expression requires special attention.In this study,it is important to emphasize the complexity of applying the traditional art concepts of form and content to software technologies.In terms of form and content,software is a textual expression with a functional basis.For copyright regime,the personal,creative contribution of the author relates only to the source code(as well as to the object code created during compilation)and its expression,while the main task of any software(whether it's AI-related or not)is primarily to execute a function.However,the unique combination of algorithms underlying the software is not protected by copyright,hence the functional component remains unprotected.Laws and regulations of the Russian Federation and the EU indicate that patenting is not applicable to software and algorithms,although patent offices continue to register software,including those created with the use of AI technologies,as software and hardware solutions.The application of the patent law to the software finds a mixed response in the judicial practice of the EU and Russia.As a result,the effectiveness of patenting becomes somewhat uncertain.In addition,the economic risks arising from patenting algorithms can lead to the destruction of competition in IT,as well as increases the risks for small companies in a collision with patent trolls.In Russia,such IT giants as Yandex Corp.,introducing neural network technologies,often refuse to patent their developments,despite this possibility,preferring to protect developments in the trade secret regime.This research tends to overview other IP systems,as well as Russian and European,especially US and Japanese IP law,intellectual property law.This is primarily due to the highly developed IP law in US,as well as the participation of US technology companies in many cases and the extensive experience courts in the development of copyright and patent law issues.In the legal systems of the Russian Federation and the EU there is no such term as artificial intelligence or neural network,although the EU in 2016 and was released draft report with recommendations to the Commission on Civil Law Rules on Robotics,but this document has no legal force,although it may affect the development of legislation on artificial intelligence in the future.The research places great emphasis on theory and doctrine,as legislators and courts of the Russian Federation and the EU in matters regarding to intellectual property often rely on legal doctrine in cases of copyright infringement on databases and code.Some scholars note that AI technologies require the development of a new doctrine,as well as major changes in legislation,but it is difficult to agree with this approach,because it is necessary to take into account the conservatism of the legislative sphere of IP law,which is still largely based on the conventions of the 19 th century.The structure of this study is related to both the structure of the contemporary AI systems and the structure of the legal regimes of intellectual property.Consolidation of technological and legal structures is reflected in the intellectual property object created during the development of the technological project.The protectability of these objects depends on the specifics of the processes of intellectual activity involved in their production,as well as their place in the legal system and interpretation during the trials.The system of Russian copyright is inherited from the European continental system,and in many ways still tries to follow it thanks to international treaties and the specifics of legal regulation in the field of IT.Although the legal systems of the EU and the Russian Federation do not belong to a common law system,the courts are forming a usual tradition,and the decisions of the Supreme Arbitration Court and the Supreme Court in the Russian Federation have recently been considered as a model for resolving similar cases.Accordingly,in this study a lot of attention is paid to the judicial interpretation of cases related to the key provisions of copyright,which are directly related to the IT industry.In conclusion it should be said that with the development and widespread introduction of technologies of AI,the questions posed to legal institutions dealing with intellectual property rights will become more challenging and it is likely that legislation on the protection of software code will have to follow the example of the protection of databases and to include the code in a sui generis legal regime,taking into account the growing potential of non-creative products of intellectual activity.However,this will become more obvious only when the development of AI technologies will affect world economic processes more.