G7 countries agreed on principles for regulating artificial intelligence

Advances in artificial intelligence (AI) technologies are making governments want to control these processes: on Monday, G7 countries adopted a statement on 11 principles for such regulation, including the introduction of a labeling system for content created using AI. At the same time, a presidential decree was published in the United States obliging developers of powerful AI systems to share the results of security testing and other important information with American authorities. Experts note that the G7 statement actually prescribes the priority of developing international regulation of AI under the leadership of the United States.

Leaders of the Group of Seven (G7) countries on Monday released a joint statement on “regulatory guidelines” for AI. The document contains 11 provisions indicating the main problems and risks posed by the development of this technology. In particular, it is proposed to introduce labeling of content created using AI so that users can identify it. Developers, according to the G7, should communicate the capabilities and limitations of AI systems, as well as areas of its “appropriate and inappropriate application.” Companies should share incident data, including with government, civil society and academia.

Yesterday, US President Joe Biden’s decree on safe AI was published, with the help of which the White House is trying to take control of the risks of this technology through more specific administrative methods. The document obliges developers of powerful AI systems to share security testing results and other important information with American authorities. It concerns companies developing AI models that could potentially pose a risk to US national and economic security, public health and safety.

Managing Director and Head of the AI ​​Regulation Center at Sberbank Andrey Neznamov notes that the G7 document is close to the AI ​​principles already adopted by the OECD and largely repeats the provisions of UNESCO declarations on AI ethics. However, there are also relatively new provisions, such as recommendations on content labeling. This is a response to the development of generative systems. According to him, the statement clearly indicates the priority of developing international regulation and standards under the leadership of the United States. The US and G7 documents contain very similar formulations on this matter – for example, they emphasize the importance of using AI to solve UN sustainable development problems. Mr. Neznamov noted that both documents were released before the AI ​​Regulation Summit, which will be held in the UK on November 1, and this looks like creating a basis for discussing AI issues at this meeting.

Alexandra Orekhovich, Director of Legal Initiatives at the Internet Initiatives Development Fund, notes that for a long time, the position of soft regulation of AI prevailed in most countries. However, recently these technologies have begun to develop at a pace at which the extent of their influence has revealed threats: invasion of privacy, discrimination, deepfakes, she notes. As a result, a number of countries have begun to think about restrictions on AI. At the same time, each took its own path: China – towards censoring content, the EU – towards protecting individuals from discrimination and social scoring technologies. “In general, like any regulation, regulation of AI will most likely slow down the development of technologies in certain, most high-risk areas. But it is obvious that the countries introducing it are calculating this risk,” notes Alexandra Orekhovich. General Director of the Agency for Transformation and Economic Development Vladislav Onishchenko believes that the adoption of regulatory documents and technology development standards will help reduce risks and increase the appetite of primarily long-term institutional investors: now the unlimited development of AI technology reduces the potential interest of investors.

Venera Petrova