Page 26 - AI for Good - Impact Report

AI for Good

The objectives of the WSC standards collaboration include providing a global platform
for dialogue on priority areas and technical standards for AI watermarking, multimedia
authenticity, and deepfake detection technologies. It aims to map the existing landscape of
standards in these fields and explore how they align with policy and regulatory requirements
for governments, ensuring transparency and compliance with legal obligations, such as privacy
protection, consumer rights, and intellectual property rights. The collaboration also seeks to
facilitate the sharing of knowledge and lessons learned from diverse stakeholders, promoting
the dissemination of emerging standards and technologies that address the challenges posed
by deepfakes and the authentication of GenAI content. Furthermore, it aims to identify gaps in
current standards and develop new standards as necessary, given the rapidly evolving nature
of GenAI and deepfake technologies.

Furthermore, ISO has established a series of standards that provide guidelines, frameworks, and
best practices for the development, deployment, and use of AI technologies. These standards
are intended to foster the safe, ethical, and effective application of AI across diverse industries.
A notable example is the ISO/IEC 22989:2022 standard, which is foundational in the AI field. It
98
provides a unified and consistent understanding of AI concepts and terminology by establishing
a specialized vocabulary that serves as the basis for related standards.

In addition, the IEC plays a significant role, particularly in the integration of AI with electrotechnical
systems. Through the ISO/IEC Joint Technical Committee 1 (JTC 1), the ISO and IEC collaborate
99
to develop critical standards for AI technologies. This joint effort encompasses both broad
and technical aspects, facilitating the seamless incorporation of AI systems. The International
Telecommunication Union (ITU) also contributes to AI standardization, particularly in the context
of telecommunications and information and communication technologies (ICT). For example,
ITU-T Y.3172 outlines how AI can enhance network performance, manage resources, and
100
optimize services, reflecting the ITU’s commitment to using AI for improved telecommunication
networks.

To ensure compliance and address complex matters such as bias, detailed definitions and
catalogues are necessary. In the medical field, efficacy is established through rigorous trials—a
stringent approach that could similarly be applied to AI systems. Moreover, AI governance
standards should be rooted in a human-centric approach that prioritizes ethical considerations
over mere effectiveness, ensuring that human rights and broader ethical implications remain
central to AI development and deployment.

European Union

On a regional level, the European Union (EU) AI Act is the most notable approach to regulating
AI – it is the first comprehensive AI regulation. It entered into force on 1 August 2024 and is now
applicable across all 27 member states of the European Union with significant extra-territorial
reach for AI providers that offer their products or services on the EU market.

16