Research On Legal Problems Of Artificial Intelligence Algorithm Infringement

Under the background of "the third wave of artificial intelligence" in which artificial intelligence algorithms are rapidly applied,the infringement problems brought by algorithms gradually appear.The closure and high speciality of the "black box" operation within the algorithm block the cognition and exploration of ordinary individuals,which makes the identification and relief of algorithm infringement an unresolved problem.Artificial intelligence algorithm takes advantage of this scientific and technological superiority to start to have control and influence on society,and human dependence on it has gradually expanded,and the "algorithm power" is quietly rising.How to regulate and relieve the infringement of artificial intelligence algorithm has become a new problem that the global society is facing and needs to solve urgently.To try to solve the above problems,the article is divided into six parts.The first part discusses "the theoretical basis of artificial intelligence algorithm infringement legal regulation".Through the analysis of the development stage of artificial intelligence,it is clear that the algorithm does not have the qualification of legal subject at this stage;As an objective problem,the judgment of its object attribute is determined by the nature of the algorithm and should belong to trade secrets,but this attribute cannot hinder the legitimacy of regulation and supervision on it.The paper seeks the legal basis to regulate the algorithm,especially its "black box",and obtains legal support for breaking through the "black box" of algorithm and exploring the root of algorithm infringement.Clarify the basic principles of artificial intelligence algorithm infringement regulation,and allow artificial intelligence technology to grow on the basis of establishing the insurmountable bottom line and ensuring the fundamental interests of mankind.The second part analyzes the difficulties of existing laws on the regulation of algorithmic infringement,and seeks a breakthrough.Due to the "legal vacuum" formed by the algorithm "black box",the tort liability subjects under the existing legal regulation are often limited to the traditional liability subjects or the algorithm application platform under the result supervision path,and the actors behind the "black box" are difficult to get negative legal evaluation under the "sunshine";The constituent elements are also difficult to identify due to the unobservability of the internal operation of the algorithm and the secrecy of the algorithm results.In order to break through the closure of the "black box" of artificial intelligence algorithm,by appropriately increasing the obligations of algorithmic benefit recipients,giving potential victims more rights,a set of rights and obligations system with algorithm as the object is set up,including data rights system,algorithm interpretation right,algorithm information disclosure obligation and algorithm review and supervision obligation.The third part determines the responsible subject of artificial intelligence algorithm infringement.The designer’s responsibility among algorithm developers should be separated from the producer’s responsibility and bear the fault responsibility independently;The responsibility of algorithmic sellers is basically consistent with that of traditional sellers;Data provider is the subject that provides data to artificial intelligence by means of operation for the purpose of profit,and data provider adopting automatic processing technology should apply fault presumption responsibility;Algorithm application platform is the medium connecting end users and developers,among which no-fault liability should be applied to public service application platform to reduce the risk of power alienation,and fault presumption liability should be applied to other application platforms;The influence of individual participation on algorithm infringement is mainly concentrated in the infringement of artificial intelligence with material entity,and fault liability should generally be applied.There is no accountability for individual participation in immaterial artificial intelligence;The algorithm infringement caused by the third party is mainly represented by hacking.If the developer fails to fulfill the security defense obligation,it is advisable to assume supplementary responsibility.The fourth part studies the constitutive requirements of tort liability of artificial intelligence algorithm.For the illegal acts,in addition to the traditional identification standards of illegal acts need to be updated,the violation of the obligation system with algorithm as the object mentioned above also belongs to the illegal act.As for the damage fact,apart from the typical personal or property damage caused by algorithmic infringement,whether the uninterest caused by algorithm "kill mature",algorithm bias and "information cocoon room" can constitute the damage in the sense of tort law is worth in-depth discussion.The criterion for determining causation should be based on considerable causality;Once the victim completes the preliminary proof of the algorithm "black box" tort causal relationship,the burden of proof should be transferred to the doer;Multi-cause and one-result is the normal state of algorithm infringement,and the responsibility distribution under the mixed causality is combed in detail.The key difference between technical fault and subjective fault lies in whether human beings have the ability to foresee and prevent the damage;The algorithm infringer should apply the more stringent fault judgment standards;The identification of the user’s takeover fault in "man-machine switching" must be reasonably determined according to the human’s reaction ability.The fifth part discusses the artificial intelligence algorithm tort liability way.Although the Civil Code separates the way of absolute right liability from the way of tort liability,the way of tort liability is not limited to damages.The non-damage compensation liability mode of algorithm infringement can be divided into preventive mode and remedial mode.Damages is still the most widely applicable liability in algorithm torts,and punitive damages should be extended in algorithm torts with a large number of victims.The sixth part advocates the construction of supporting system.Aiming at the problem that the number of victims of algorithm infringement may be huge,by analyzing the shortcomings of the civil public interest litigation and joint litigation system in China,it is advocated to build a class action system of algorithm infringement;Considering the current level of algorithm development,artificial intelligence compulsory liability insurance should build a dual-track insurance system for owners and developers with "man-machine switching" as the demarcation point;In addition,an algorithm infringement relief fund should be set up to provide timely assistance to victims when conditions are met.