G7 agrees on first comprehensive guidelines for generative AI
By Mayumi Hirosawa
TOKYO — Group of Seven digital and technology ministers agreed Friday on the world’s first comprehensive international guidelines for generative artificial intelligence, applying to both developers and users, to address such problems as misinformation.
The principles sit alongside a voluntary code of conduct for AI developers agreed on by G7 leaders in October as the two centerpieces of the group’s Comprehensive Policy Framework. The guidelines are expected to be approved in a virtual summit as soon as early December.
Of the 12 points in the new principles, 11 are drawn from the code of conduct, applying them to users of AI services as well as suppliers of them.
The new 12th principle is to “promote and contribute to trustworthy and responsible use of advanced AI systems,” such as improving digital literacy and sharing information on risks and vulnerabilities. It specifically notes risks related to disinformation.
Advanced AI can generate convincing fake images and video, contributing to the spread of misinformation on social media. There are concerns among G7 members about these capabilities being abused, including by such state actors as China and Russia.
The guidelines also call for developing and adopting reliable authentication technology, such as digital watermarking to identify AI-generated content, and implementing protections for personal information and intellectual property.
The document is part of the final stage of the Hiroshima AI Process, a forum launched in May under Japan’s G7 presidency to discuss international rules for advanced AI systems. The Japan-led process aims to set a de facto international standard by leading the way on setting comprehensive rules.
The G7 ministers agreed on an implementation plan for next year — including the development of monitoring tools to track compliance with the principles — and outreach to encourage other countries to sign on.
They also discussed the concept of “data free flow with trust” to facilitate cross-border data transfers, which can be hindered by regulatory differences between countries.
The group agreed to establish an international body under the umbrella of the Organisation for Economic Co-operation and Development to discuss topics including privacy-enhancing technologies like encryption.
