Artificial Intelligence 'Desdemona' (R) by Yaya Labs is part of a presentation during the International Consumer Electronics Fair (Internationale Funkausstellung Berlin, IFA) in Berlin, Germany, 01 September 2023. EFE/EPA/FILE/CLEMENS BILAN

G7 agrees on code of conduct for AI developers

Tokyo, Oct 30 (EFE).- The G7 countries on Monday agreed on a code of conduct for companies and institutions that develop artificial intelligence (AI) systems, aimed at mitigating risks derived from this technology such as disinformation and the violation of privacy or intellectual property.

The leaders of the Group of Seven – made up of Germany, Canada, the United States, France, Italy, Japan and the United Kingdom – expressed agreement on guidelines for the AI sector, dubbed the “Hiroshima Artificial Intelligence Process,” according to the Japanese presidency.

This code of conduct seeks to promote the development of “secure and trustworthy” AI systems internationally and “address the risks” associated with the technology, according to the joint statement, which also urged the AI industry actors to commit to compliance.

The G7 underlined the innovative and transformative potential of advanced AI systems, and in particular generative models such as the ChatGPT conversation bot, while recognizing the need to protect individuals, society and shared principles, while keeping humanity at the center.

In this regard, the G7 launched last May at the summit in Hiroshima (west) an initiative that has led to the elaboration of 11 principles for companies and organizations involved in the development of AI systems.

Recommended measures include monitoring by independent observers at all stages of AI development and market entry, and ensuring that AI creators identify potential risks of misuse and address these vulnerabilities.

In order to increase transparency of these models, developers are asked to publicly report on the “capabilities and limitations” and domains of appropriate and inappropriate use of their systems.

It is also called to implement authentication mechanisms such as the creation of digital watermarks that allow users to identify texts, images or videos generated by artificial intelligence.

These guidelines are to be reviewed periodically when necessary to adapt to the “rapidly evolving technology” and in consultation with stakeholders, according to the joint statement.

Besides urging AI developers to follow this code of conduct, G7 countries agreed to accelerate the process to develop a policy framework to promote implementation of the Hiroshima process, including project-based cooperation.

The G7 will also promote compliance with this code of conduct through cooperation and consultation with different international organizations such as the Global Partnership on Artificial Intelligence, the Organisation for Economic Co-operation and Development and with public and private actors, as well as countries that are not part of the G7. EFE