Russian and foreign regulation processes in 2020 received a strong impetus for development: American, European and Russian regulators published and approved new documents, standards and data in the artificial intelligence (AI) field, countries formulated new global goals and objectives for the development of the technology.

ICT.Moscow talked with experts in the AI regulation, summed up the results of 2020 and found out about the possible vectors of development in 2021. The experts were: Executive Director of the Data Research Center for Government Bodies of Sberbank Andrey Neznamov, Director of the Department of Strategic Development and Innovation of the Ministry of Economic Development of Russia Rustam Tikhonov, Chairman of the Technical Committee (TC) for standardisation 164 “Artificial Intelligence” Sergey Garbuk, Head of Artificial Intelligence at the IT Department of Moscow Government Ivan Buturlin, Deputy Department Director for the Regulation of Cyberphysical Systems at Skolkovo Foundation Alexander Tyulkanov, spokesman of the working group of NTI “Autonet” Yaroslav Fedoseev.

Shift from developing a strategy to detailed regulation

By the end of 2019, at least 37 countries of the world have already formulated their own strategies or plans for the development of AI, and the plans of another 13 countries were under development, according to the OECD.

Last map update was made on November 22, 2019

In 2020 Saudi Arabia joined the countries that have developed their national AI strategies. In October, Saudi Arabia launched a strategy up to 2030, announcing plans to enter the top 15 world leaders in AI by that time. And in the same month the USA approved the strategy of world leadership in the area of developing technologies, including AI and neural interfaces.

Executive Director of the Data Research Center for Government Bodies of Sberbank Andrey Neznamov notes that the principle of developing strategies for artificial intelligence is mostly a closed chapter, and in 2020 world regulators took the next step. Now they are developing detailed approaches to regulation: in particular, they are developing documents that determine vectors for regulation for the next few years. It has become one of the key trends in the technology regulation in the world.

We are witnessing a change in the overall trend in regulation. Until 2020, the approaches to AI and its regulation were strategized, and every national strategy had a part on regulation and ethics. In some strategies it is a reference (paragraph or line), in other strategies - a section, and in some strategies it is a driving factor, as it is in Russia.

In 2020, we observed the next stage, when experts started developing detailed regulation - not the acts themselves were adopted, but approaches towards the regulation of AI were being thoroughly developed. We see that different countries in the world have developed documents that shape vectors in regulation for the next five years.

Andrey Neznamov

Executive Director of the Data Research Center for Government Bodies of Sberbank

This trend is also mentioned by Ivan Buturlin, Head of Artificial Intelligence at the IT Department of Moscow.

Many countries continue to create national concepts of artificial intelligence, megacities are creating strategies for the development of AI within the framework of smart cities. Thorough documents that contain specific approaches to the legal regulation of AI are also being developed. In general, regulators are paying increased attention to digital technologies and the Internet.

Ivan Buturlin

Head of Artificial Intelligence at the IT Department of Moscow Government

Speaking about the most significant documents that shape the vectors in the area of regulation for the coming years, Andrey Neznamov names three regions where the largest number of significant documents in the global regulation of AI and related areas were created in 2020:

1. In the United States - a presidential executive order on the promotion and use of trustworthy AI in the federal government, adopted on December 5 (the draft was published at the beginning of the year). (UPD 22.01.2021: the document was later removed from the US White House website)

2. In the European Union - the “White Paper on Artificial Intelligence”, a document which was published in February (draft version), and which defines a large number of approaches to certain problems of AI: what requirements to establish, how to control AI, etc.

3. In Russia - the Concept for the Regulation of AI and Robotics until 2024, approved by the government in August 2020. It defines both general regulation principles and industry approaches.

Intensification of international cooperation

The second noticeable trend in global regulation which the experts mentioned is the intensive work of international authorities. This trend was also mentioned by the Director of the Department of Strategic Development and Innovation of the Ministry of Economic Development of Russia, the agency responsible for coordinating activities for the development of artificial intelligence ecosystem.

There have been many developments in the world regulation system this year. The first document on AI ethics - Recommendation on the ethics of Artificial Intelligence - is being developed at UNESCO. The Council of Europe is developing a document in the field of human rights in AI. Russia is actively involved in their development. The main global trends are shaped by the fact that more and more sites and platforms start to form a global position on AI.

Rustam Tikhonov

Director of the Department of Strategic Development and Innovation of the Ministry of Economic Development of Russia

Andrey Neznamov highlighted three global communities whose work was most noticeable in 2020 and shared some details of their work:

1. The Council of Europe and Ad hoc Committee on Artificial Intelligence (CAHAI) are actively developing approaches to European regulation, and in 2021 it will start to conduct consultation on these approaches with the member countries of the Council of Europe.

2. UNESCO, which developed the draft recommendation on ethics of artificial intelligence. Member countries of UNESCO will discuss this document in 2021.

3. The Organisation for Economic Cooperation and Development (OECD) launched AI ​​Policy Observatory in March. It comprised information about AI regulation around the world.

Earlier, Alexander Tyulkanov, Deputy Department Director for the Regulation of Cyberphysical Systems at Skolkovo Foundation, also told ICT.Moscow that “some problems, including standardisation in the area of artificial intelligence technologies and harmonisation of approaches to humanitarian-oriented regulation of the use of these technologies can only be solved together, within such international organisations as ISO, UNESCO, the Council of Europe”. He believes that it is possible to take into account the interests of Russian developers in the process of development of regulatory documents by such organisations.

However, sometimes the aspirations of such coalitions not necessarily correspond to the interests of individual counties.

Judging by what we see, this regulation may be not in line with the national interests. It often happens with international conventions, but in the case of AI, this is an especially sensitive issue.

Artificial intelligence is a general concept, and we need to talk about specific interpretations and applications of AI in each area. Each area will have its own regulations. And now there are attempts to create a certain general template and do everything according to it. This is the dead-end road, and if we follow it, we can destroy the industry instead of formulating the rules of the game.

We also need to understand that countries have different levels of maturity in the use of technologies, different challenges and cultural paradigms, as well as different legal systems. It seems to me that working out general rules for AI at the global scale is a very difficult but necessary task.

Andrey Neznamov

Executive Director of the Data Research Center for Government Bodies of Sberbank

The expert also mentioned the statements about the global race in artificial intelligence, in which China and the United States are leading. However, from his point of view, creating regulations is only part of this race to build a comfortable legal framework for the development of the technology and for developers.

For instance, in China, according to Andrey Neznamov, there is “a very peculiar regulatory system, but at the same time an environment that is quite comfortable for artificial intelligence technologies has been created. However, the information about what they are doing is often closed from free access. I suppose that China will go its own way and will not set a strict framework for AI through regulatory measures”.

The expert also named the Global Partnership on AI announced in May, in which the OECD countries became participants, as one of the key results of the year. However, neither China nor Russia, whose parliaments agreed to intensify the exchange of experience in legal regulation in various areas (including regulation of the digital economy and AI) in November, are not yet part of this alliance. However, it cannot be claimed that Russia has been unnoticeable in the global AI regulation landscape.

Russia stands out on the global scale

Speaking about the most significant events in Russian regulation, Andrey Neznamov mentioned the government’s approval of the Concept for the Regulation of AI and Robotics. According to the expert, “Russia was able to get ahead most countries in the world in terms of the thoroughness of development and the speed of adoption of regulations. Almost no country in the world has such detailed regulatory concepts that would be discussed by the state with the business and scientific communities and would eventually lead to the creation of a large resulting document”.

The emergence of the first legislative initiatives directly related to artificial intelligence is also important. This is, in particular, the law on an experimental legal regime in the field of AI in Moscow. Andrey Neznamov also reminds that there is another federal initiative: the law on experimental legal regimes in the area of digital innovation, in which AI is declared as one of the key end-to-end technologies. For many experts interviewed by ICT.Moscow, including the experts from other IT industries, AI is associated with this law.

Andrey Neznamov explains, that in the case of regulatory sandboxes these two approaches (territorial and industry-specific) are common practice in the world. He adds that “there are two basic models: the case of the law with experiment in Moscow demonstrates a territorial model (let us call it ‘horizontal’), and the law on experimental legal regimes - a vertical model, which encompasses certain areas. The main thing is to have as many sandboxes as possible, because any innovation, especially AI, requires special regulation. So, the creation of such sandboxes is another key trend in AI regulation”.

AI experiments in Moscow

The first effects of AI sandboxes introduction in Russia can be assessed soon. One of them, the Moscow AI experiment, started in the middle of the year on July 1. Ivan Buturlin, a representative of the IT Department of Moscow Government (DIT of Moscow), says that the launch is extremely important for the entire industry - not only for the city’s industry, but also for the whole Russian industry.

The start of the experiment (adoption of the Federal Law 123-FZ) is a huge step towards the creation of a flexible mechanism for the implementation of AI in various areas of the urban economy. We determined what city data could be useful for retraining AI models, and identified the legal barriers that hinder the implementation of AI solutions, and also started collecting requests from segments of urban economy for AI solutions.

We expect that next year we will see examples of the implementation of such experiments, and will be able to evaluate the first results of the introduction and creation of the new regulation.

Ivan Buturlin

Head of Artificial Intelligence at the IT Department of Moscow Government

Since February, another experiment has been carried out in Moscow in the field of healthcare. It is a remarkable example of urban regulation of a narrow industry segment (AI in medicine), which has shown its effectiveness during the pandemic.

For example, back in July, Anna Meshcheryakova, CEO of Third ​​Opinion, while participating in the experiment, told ICT.Moscow about the opportunities provided to the industry participants by this initiative:

The experimental legal regime makes it possible to use innovative services if they are not specified in the procedure of medical aid provision, clinical guidelines and standards of medical care. Within the legal experiment, there is no mandatory requirement for digital medical services [...] to have a registration certificate for a medical device issued by Roszdravnadzor (Federal Service for Supervision in Healthcare).

Anna Meshcheryakova

CEO of Third ​​Opinion

Summing up the results of the year in healthcare for ICT.Moscow, Sergey Morozov also calls the Moscow experiment in healthcare the main technological event of the year.

However, the example of healthcare shows that the regulation of artificial intelligence is much more demonstrative in certain industries. The medical area was noticeable not only in Moscow, but also at the federal regulatory level, while the autonomous transport industry is lagging far behind in terms of the results achieved by the end of 2020 in AI regulation.

Two extremes in Russian AI regulation

Andrey Neznamov evaluates the development of legal regulation of AI in medicine extremely positively. This area is developing at a breakthrough pace, and it is “not only synchronized with international regulation, but even outpaces it”.

Healthcare is an every sensitive segment, the introduction of new technologies in it should be as safe as possible, therefore, the regulation in this area is traditionally conservative for the safety of the people. However, in health care the regulator and the business managed to achieve biggest results and find a balance.

Andrey Neznamov

Executive Director of the Data Research Center for Government Bodies of Sberbank

Among the most notable events in this area, the expert mentions the following:

• At the end of the year, an amendment to the government decree on the approval of the rules for state registration of medical devices was adopted. The registration procedure for AI systems, and for medical software in general, has been simplified. Also in February of 2020 criteria for classifying software as medical devices were developed.

• In the middle of the year the issue of the classification of AI systems into hazard classes was solved. They were classified as the third class (“high-risk software”) - AI systems are considered the most dangerous class of medical devices, until special rules for registering AI systems are developed. “To create such rules we need special rules for conducting clinical trials, checking datasets, etc. The drafts of such rules have been prepared as technical standards for medical systems (they are currently under discussion)”. 

• Also in the middle of the year the rules for conducting X-ray examinations were developed, and these rules are indirectly related to AI: they mention medical decision support systems.

In April, Anna Meshcheryakova, CEO of Third Opinion, an AI-based healthcare company, told ICT.Moscow that regulatory mechanisms helped “to accelerate the development of the industry and increase the number of implemented solutions”. Practice confirms it: for instance, in 2020 registration of AI-based medical software really was accelerated. Three such solutions have already been registered: the first was Webiomed (in April), by the end of the year Botkin.AI (in November) and Care Mentor AI (in December) joined it.

Another area - autonomous vehicles - on the contrary, was assessed rather negatively. According to Andrey Neznamov, it was not possible to achieve significant results in this segment during the whole year, which exacerbated the lagging behind global trends in regulation.

On April 1, the President ordered to approve a set of measures for testing and putting into operation autonomous vehicles, including without the presence of an engineer in the vehicle, by May. To prepare a set of actions, an informal consortium of all unmanned vehicle manufacturers in Russia was created (KAMAZ, Yandex, Sberbank, Gazprom Neft as one of the key customers, small startups and laboratories). Business proposals were prepared and sent to the government.

Unfortunately, no changes have been made to the decree that regulates the testing of unmanned vehicles yet. The business has also prepared its proposals for changing the regulation, which are synchronized with the draft of the general set of actions in the field on autonomous vehicles.

Finally, at the beginning of the year, a draft law on innovative vehicles was introduced to the State Duma (already the second draft). Like in the case with the first one, it was returned for revision.

Andrey Neznamov

Executive Director of the Data Research Center for Government Bodies of Sberbank

The representative of Yandex Self-Driving Technologies also notes: despite the fact that the technology is already ready for launch, the implementation is hindered by the regulatory barrier that leads to Russia’s lagging behind competitor countries. According to the representative, the most significant launches of autonomous vehicles in 2020 took place in the United States and China, where legal framework for this has already been created: in Phoenix in the USA (a robotic taxi service that operates without a safety driver) and in several Chinese cities.

However, Yaroslav Fedoseev, the representative of NTI Autonet, mentions some positive shifts that have occurred in 2020.

This year Autodata project was discussed; it describes the algorithm for collecting data from the vehicles. This is very important for the autonomous driving industry, because today this data can be stolen or subject to cyberattacks. The bill will help secure this data, create new services and technological developments on its basis.

Yaroslav Fedoseev

Representative of NTI Autonet

We cannot say that absolutely nothing happened in the regulation of self-driving vehicles in 2020. But the adopted measures are clearly not enough to keep Russia at the level of other countries that are actively developing the regulation in the field of autonomous cars. However, unmanned vehicles are not the only challenge in the domestic regulation.

Work with data

According to Andrey Neznamov, there still is uncertainty in data regulation in Russia, and it hampers the development of artificial intelligence.

Data regulation is critical in developing AI technology. The clearer and more consistent the data access rules for developers, the better.

However, there still are several problems. For example, the business believes that it is important to introduce into Russian law the data anonymization institution, where data is irreversibly depersonalized, and no connection can be established with a specific subject. In July, a draft on amending the law on personal data was introduced to the State Duma. The draft has been initially developed for several years to clarify this issue. But after discussions the provisions on anonymization were excluded from the draft submitted to the State Duma.

This is a problem, because, according to the National strategy on artificial intelligence, data is very important for AI technologies, and it is necessary to create special access to data for the developers. There is little progress here. A draft on providing such conditions for AI developers that helps to remove unnecessary barriers is currently being discussed. In addition, this year we have prepared a draft of the methodology for anonymizing personal data (it is considered one of the biggest problems) and offered to carry out a pilot project in Moscow.

Andrey Neznamov

Executive Director of the Data Research Center for Government Bodies of Sberbank

However, there was a clear demand for regulatory measures on the market in 2020 - it was noted by industry representatives with whom ICT.Moscow spoke in April. For instance, Ivan Begtin, chairman of the Association of Data Market Participants, said that it was necessary to regulate the disclosure of data by the state, since it will contribute to the development of the AI ​​market. And according to Alexander Krainov, head of the Yandex Machine Intelligence Laboratory, in the process of data regulation it necessary to find a balance between protecting user information and collecting data for the development of technologies and services, and this process should be flexible.

In any case, it cannot be claimed that everything is bad in the field of data. Rustam Tikhonov from the Ministry of Economic Development told ICT.Moscow about one of the most prominent achievements in this area in 2020.

This year we have approved a standardisation program that includes the development of over 200 standards in the field of artificial intelligence. I would like to say that they were developed not to create problems for business, but to standardise business approaches to the development of AI, reduce costs, ensure the safety and rights of citizens when using these technologies.

Rustam Tikhonov

Director of the Department of Strategic Development and Innovation of the Ministry of Economic Development of Russia

Chairman of the Technical Committee (TC) for standardisation 164 “Artificial Intelligence” Sergey Garbuk also speaks about the positive data initiatives that were implemented in Russia in 2020, and shares some details about the standardisation program.

There are two major developments in the ​​national standardisation. One of the is the adoption of a number of standards in the field of AI and big data (Autodata Platform standards, a set of national standards that are in the final stage of elaboration), and the other is the approval of the perspective standardisation program in the priority area of ​​Artificial Intelligence in 2021-2024. The program includes 217 general purpose and industry metrology standards and was created to support the development and application of AI in such industries as public administration, healthcare, transportation, natural resource management, industry, education, construction, telecommunications, trade, prevention and remediation activities.

Sergey Garbuk

Chairman of the Technical Committee for standardisation 164 “Artificial Intelligence”

Alexander Tyulkanov, Deputy Department director for the regulation of cyberphysical systems at Skolkovo Foundation, agrees with his colleagues. He says that one of the key events of the year was the government’s approval of the Concept for the regulation of AI and robotics, with the help of which the market received guidelines for developing regulation of this area in Russia, and he also mentions the issue of access to data.

The government agreed that our developers need special conditions to access data, including anonymized personal data, on condition that the measures to protect the interests of citizens are taken. There is no artificial intelligence without datasets for training.

I would also like to emphasize that, of course, we are not talking about the legalization of any sale of personal data or arbitrary use of data. The opportunities for using data for research purposes, providing machine learning in the so-called data sandboxes are currently being created.

Alexander Tyulkanov 

Deputy Department Director for the Regulation of Cyberphysical Systems at Skolkovo Foundation

It should also be mentioned that in 2020 one of the notable events was the creation of open city AI registers in Amsterdam and Helsinki. Ivan Buturlin from the IT Department of Moscow Government told ICT.Moscow how he assesses this practice and whether Moscow plans to create open city AI datasets.

Such platforms help increase city residents’ confidence in AI systems, improve their quality through feedback from the professional community, and reduce risks during development, training and implementation of AI projects. We regularly study the best international practices and search for successful practices for implementation in Moscow, such as the practices implemented in Amsterdam and Helsinki.

We understand that the ability to use urban transactional datasets will become an important driver for the industry. We are already forming urban datasets that we can offer to participants of the Moscow experimental legal regime.

Ivan Buturlin

Head of Artificial Intelligence at the IT Department of Moscow Government

Despite all the difficulties, in general, experts positively assess the prospects of Russian legislation: they expect that law on experimental legal regimes in the area of digital innovation will be adopted, and that the problems with data and autonomous vehicles will be solved soon.

Expectations from 2021

Sergey Garbuk hopes that “in 2021 and the years to come the role of the leading domestic developers of AI technologies as drivers of standardization in this field will increase”.

Firstly, companies will be motivated by the desire to overcome regulatory barriers associated with the need to provide the guarantees of security of AI systems to the public (for example, in transport, healthcare and other industries), resolve the regulatory conflict between information security requirements and providing access to thematic datasets for developers. Secondly, they will be motivated by the interest in creating uniform principles for measuring the characteristics of AI systems that will help to objectively compare similar-purpose systems from different developers.

Sergey Garbuk

Chairman of the Technical Committee for standardisation 164 “Artificial Intelligence”

Andrey Neznamov believes that data standardisation will become one of the key trends in Russia, about which people will be talking with confidence in a year.

Technical regulation is designed to determine specific technical requirements for AI systems; at the same time, it is a kind of alternative to (or supplement) to legal regulation. Those details that are impossible and impractical to capture in legislative acts should be regulated by GOSTs or technical regulations.

The global system for AI standardization is being created right now, and Russia has not been left out. So far, several standards have been published, but this is very little compared to the number of standards that will be approved next year. This work has just begun. Several dozen standards in individual areas were submitted for discussion. I suppose that key results in this area will be available in 2021-2022, when it will be possible to demonstrate what standards have been developed, approved and what they were created for.

Andrey Neznamov

Executive Director of the Data Research Center for Government Bodies of Sberbank

The expert believes that Russia has defined the conceptual problems and the ways to tackle them, and now it is necessary to work out individual industries, first of all, autonomous vehicles and data, but it is still too early to make any forecasts.

Ivan Buturlin from the DIT of Moscow is looking forward to getting the first results of experimental legal regimes, and also expects that regulation will be further developed within “Artificial Intelligence” federal project, which was approved in August.

We can expect the emergence of new regulation in providing support for the companies developing AI solutions, standardisation, incentives for the development of AI, and also in the development of supplementary vocational education.

Ivan Buturlin

Head of Artificial Intelligence at the IT Department of Moscow Government

Rustam Tikhonov from the Ministry of Economic Development forecasts that the main trend in AI in 2021, at least in Russia, will be the launch of large-scale regulation and standardization. He told ICT.Moscow about some plans for regulating artificial intelligence.

We plan to develop a roadmap for transforming the business climate within Artificial Intelligence area. The document will change the regulation in the field of data, shape the regulation of AI in transport and healthcare, clarify responsibility issues, etc. These regulations will be developed with the involvement of a large number of business representatives. We hope that this will make it possible to form a regulation that minimally interferes with business, protects the rights of citizens and at the same time provides great opportunities for the development of AI in the country. The roadmap will be supplemented with new events, including the events in other sectors, every year.

Rustam Tikhonov

Director of the Department of Strategic Development and Innovation of the Ministry of Economic Development of Russia

Also the representative of the agency says that the Ministry of Economic Development has prepared "a project passport within the framework of the National Strategy for the Development of AI, which was approved by the President of Russia in October 2019. According to the project, the main activities are planned for 2021, we are not falling behind the schedule”.

Thus, all indications are that legal regulation of artificial intelligence in 2021 will be actively developing: in Russia, as elsewhere in the world, this technology will be one of the priority areas.

Original (in Russian)