Historic G7 agreement on Artificial Intelligence

The G7, an intergovernmental political forum made up of Germany, Canada, France, the United States, Italy, Japan and the United Kingdom, reached an agreement on the international guiding principles of artificial intelligence, as well as a voluntary code of conduct for companies that develop advanced artificial intelligence systems within the framework of the Hiroshima AI process.

This voluntary code of conduct represents a milestone in the way these countries govern AI. This technology has raised concerns about privacy and security risks, beyond the unquestionable opportunities it offers.

Last May, the political leaders of the countries that make up the G7, together with the European Union, began the process in a ministerial forum called “Hiroshima”. AI process”.

The code, which consists of eleven points, seeks to promote safe and reliable AI, providing voluntary guidance to those entities that develop advanced AI systems, including basic models and generative AI systems.

In this sense, the code intends that those companies involved take appropriate measures to identify, evaluate and mitigate risks throughout the life cycle of AI. In the same way, the code urges these companies to address incidents and patterns of misuse after the commercialization of the products.

Companies that voluntarily subscribe to the code will have to publish public reports on capabilities, limitations, and the use and misuse of AI systems. On the other hand, companies will need to invest in strong security controls.

In this context, the third way has been imposed, that alternative between the position of the European Union and that of the United States. Thus, the alternative proposed by Japan has been posited as an intermediate point between the other two. Specifically, the government of the Japanese country brought to the table a looser regulation than the European one, but specific enough to address problems arising from the use of works protected by intellectual property rights or the exposure of personal data.