The production model of Internet content has undergone many changes，After the professional generation content and user generation content，Now it has entered the production method of automatic or assisted content through artificial intelligence technology, that is, the era of artificial intelligence generation。Recent，Several new types of new generation artificial intelligence applications are popular，Known as the most revolutionary technological progress since 1980，It indicates the arrival of the fourth industrial revolution。but，Where is the benefit，Faculture is also followed。The potential social risks of generating artificial intelligence began to be revealed，It is data security、Privacy risks and algorithm security have aroused widespread attention。Germany、France、European countries such as Italy began to consider adopting stricter supervision，Regulatory storm for generating artificial intelligence has arrived。2021，The EU proposed the draft of the Artificial Intelligence Act，The purpose is to provide resolving artificial intelligence risks、Basic legal framework for protecting EU citizens' rights。July 10, 2023，Seven departments such as the National Network Information Office of my country review and approve the "Interim Measures for Management of Genesis Artificial Intelligence Services" (hereinafter referred to as "Interim Measures")，It aims to "stand rules" for generating artificial intelligence。It can be expected is，The regulatory system for generating artificial intelligence will become stricter and stricter。but，Overall，The regulatory legislation of the current countries has not prepared for the emergence of generation of artificial intelligence，Regulatory plans generally exist in a certain misunderstanding，Urgently need to clarify and Kuangzheng。

bet365 live casino games  

bet365 live casino games Genetic artificial intelligence exists in data leak、false information、Intellectual property rights and other risks，This is not controversial。The controversial is，Who and which risks are responsible for? Given the chain of production of artificial intelligence from production to publishing，The establishment of the responsibility framework should enable the affected person to apply for responsibility in the entire responsibility chain，Therefore, a unified framework of responsibility needs to be established。Although there are legal attempts to stipulate responsibility，But because the legislative concept is not updated in time，Often scattered and not a system。For example，The European Commission passed two proposals in 2022，that is, "Artificial Intelligence Responsibility Instruction" and the revised "Product Responsibility Instruction"，Therefore, the cornerstone of the EU artificial intelligence supervision。but，"Artificial Intelligence Responsibility Instruction" is based on the responsibility of negligence，"Product Responsibility Instruction" is based on strict responsibility，and the two proposals are complementary to the content of the draft of the "Artificial Intelligence Act"，It does not form a harmonious and unified responsibility framework。The relevant regulations of our country also have similar problems: more emphasizes the administrative responsibility and criminal responsibility of developers and providers，Not involved civil infringement liability，There is no principle of responsibility for responsibility，The damage to the possible rights and interests caused by the generated artificial intelligence needs to be resorted to the "Civil Code"、Law of Intellectual Property Law and other legal departments，and the victim faces great difficulties when proved the fault of developers and providers。

Misunderstanding 2: Pay attention to the current risk of generating artificial intelligence rather than long -term risks 

Although the discussion of artificial intelligence risks is in full swing bet365 best casino games now，But the legislative action is very slow。The United States posted the "Act of Artificial Intelligence Rights Act"，Among them, "Algorithm discrimination protection"，But the United States has not yet passed a law that regulates the generated artificial intelligence。The laws of the previous EU have not focused on the issue of artificial intelligence malicious use。my country's "Interim Measures" is the first regulation in the world to really pay attention to the risk of malicious use of artificial intelligence。but，Whether it is the current discussion or legal regulation，All are dealing with the current prominent algorithm discrimination、Privacy and Responsibility "Gray Rhino" problem，Lack of understanding and concern for the "black swan" problem that may occur in the long run and the future。We must understand clearly，Our understanding of the generation artificial intelligence ability，It is limited to the part of our active test and research，Generating artificial intelligence practical ability may be far more than we imagine，Among them, maybe unknown risks。Prevent unknown risks caused by unknown ability，It should be a problem that every generic artificial intelligence regulatory system must be considered carefully。

Noim Coom in the article "Algorithm Black Swan"，Hidden risks of generating artificial intelligence are divided into three categories: the first category is general risk，The generated artificial intelligence system is generally used in society，Once ethical and security problems occur in the high risk field, it will cause catastrophic consequences; the second category is the risk of diffusion and improper use，illegal acts such as bet365 live casino games Genetic artificial intelligence can be easily transformed for illegal criminal acts such as cyber attacks or financial fraud; the third category is systemic risk，生 ingredients artificial intelligence may make the public judge failure、Values ​​are negatively affected，Then it brings challenges to political and ideological security。So，We need a "preparation bet365 Play online games on the algorithm" to guide the improvement of the regulatory system，Preventing possible algorithms in the future "Black Swan"。

Misunderstandings 3: The focus of supervision is not placed at the application level but pre -training 

The existing legislation will focus directly on the risks that the generatory artificial intelligence itself may bring，Emphasize intervention at the pre -training level to ensure the generation of results that meet the regulatory requirements。but，Research Digesty，Focus on the application of applications, especially high -risk applications, may be more effective。Generating artificial intelligence model itself has high degree of degeneration and generalability，Developers may not be able to fully evaluate and foresee high risks that may bring。Many developers have such foreseeable possibilities: Once the generated artificial intelligence developed by itself，It is very likely to be used in high -risk projects。Since it is unable to exclude high -risk applications in technology，So the generated artificial intelligence can easily be regarded as a high -risk system，Developers will always be in the risk of blame in the existing regulatory mode，This is not beneficial to the healthy development of the generated artificial intelligence。Brief words，The existing regulatory model proposes a "impossible task" for developers。So，Place the supervision focus on the pre -training level，It is too heavy for developers and unnecessary，Instead, it should be placed at the application of the application, especially the high -risk application level。

Of course，There are two important exceptions to conversion of regulatory focus: First, non -discriminatoryness on the algorithm，This is the problem that should be solved at the pre -training level; the second is data protection，The data used for pre -training should meet the compliance requirements of data security and personal information protection。

Kuangzhengyi: Establish a unified framework for generating artificial intelligence liability 

Responsibility for generating artificial intelligence，Bet365 lotto review It should be stipulated in a unified legislation，to avoid the difficulties of mutual conflict between different laws and the difficulty of proof of the victim。The main point of legislation is how to balance the relationship between scientific and technological innovation and rights protection。For this，There are two problems that need attention: First, reasonable allocation of blame for blame，The boundary and strength of the protection of rights protection with reasonable determination。Damage caused by illegal development and applications and banned high -risk applications，The principle of strict responsibility can be considered; damage caused by legal development and applications and high -risk applications，You can consider applying the principle of fault and settlement; for the damage suffered by users who use generate artificial intelligence，You can consider applying the principle of fault。Second is to leave the necessary public space for the sustainability of scientific and technological innovation。The needs of scientific and technological innovation and industrial development should be fully considered，Also through security port、Regulatory sandbox and other systems to protect developers' innovative enthusiasm，The innovative protection of the bet365 live casino games Genetic artificial intelligence developer through the principle of proportion and the dynamic balance of the protection of the victims。

Kuangzheng 2: Pay attention to the long -term risk of generating artificial intelligence 

On the one hand，Legislation cannot be achieved overnight，Instead, it is necessary to continuously monitor the upgrade iteration of generating artificial intelligence，Continue to evaluate its possible risks。Legislation needs to prevent large -scale social hazards that may occur in the future of artificial intelligence in the future，Always adhere to the openness of cognition，Realize "Learning in Supervision"，Maintain enough flexibility to adapt to changes in the new situation，and consider the evolution characteristics related to emerging technologies and its application，Design Bet365 lotto review is more adaptable、Prospective regulatory framework。On the other hand，Legislation must maintain sufficient standardized elasticity，It is required that legislation can fully foresee the possible risk is difficult to occur.。Due to the complexity of the legislative procedure，Require legislation to modify it at any time to meet the needs of new situations，It is also difficult for it。So，It should be considered to establish a flexible regulatory framework，Good at using blank specifications and general nature、General clauses leave space for future supervision。Of course，In this process, the relationship between the clarity and vagueness of the law must be balanced。

Kuangzheng 3: Place the supervision focus on the application level 

Supervisor must know，It is impossible to force developers who are required to generate artificial intelligence to establish a omnipotent、The risk management system that can prevent any risk management that may bring in the future。So，Supervision of generating artificial intelligence shall transfer the focus to the application level，At least the level of high -risk application。The ideal approach is，For high -risk applications (for example，Generate sensitive content、Social and economic decision -making consultation, etc.)，Developers (providers) establish a cooperative relationship with the authorized users，Share the necessary information with each other，Developers (providers) assume responsibility for reviewing whether they are authorized，And the user assumes responsibility for improper use。Legislation shall establish a minimum compliance framework for this。Of course，This cooperation may involve national secrets and intelligence、Protection of important information such as business secrets and intellectual property rights。This needs to modify relevant laws，To form a coordinated regulatory framework，At the same time, consider the possibility of the intervention of third -party neutral agencies。

(Author is an associate researcher at the Institute of Applied Law of China)