Page 31 - AI for Good - Impact Report

AI for Good

Interoperability and Technical Standards: The Foundation for Global
AI Governance

As AI continues to evolve at an unprecedented pace, the need for a cohesive and unified
global governance framework becomes increasingly urgent. Various approaches – ranging from
the UN’s ethical guidelines, the G7’s voluntary code of conduct, the practical safety measures
initiated at the AI Safety Summits, and the OECD’s definition of AI as well as the work by the
standard-setting bodies – provide a foundation. However, the future of AI governance will
largely hinge on the ability to ensure interoperability across these frameworks. The development
of AI policies, including the creation of frameworks and the standardization of AI, are deeply
interconnected, with each process reinforcing and complementing the other. Technical
standards provide the essential specifications and best practices that form the backbone of
robust regulatory and policy frameworks.
Interoperability refers to the capacity of different AI governance frameworks to work together
seamlessly, despite diverse legal, cultural, and technical contexts. This is critical for several
reasons. First, it enables the sharing of best practices, technical standards, and ethical guidelines,
fostering a collaborative approach that includes all stakeholders – governments, industry, civil
society, and international organizations. Therefore, the application of technical standards is vital
for the safe application of AI. Learning from past standard-setting experiences, such as those in
telecommunications and postal services, can be highly beneficial. These models were founded
on values like fair market practices, competition principles, trust, and transparency – values that
are equally vital in AI governance. However, standards must be trustworthy and not rushed
to maintain their credibility. This requires detailed definitions and catalogues, particularly for
complex challenges like bias, to ensure effective compliance checking.

Second, interoperability allows for consistency in applying AI regulations across jurisdictions,
reducing the complexity and costs for internationally active organizations and businesses.
Interoperability is moreover about fostering a shared understanding and commitment among
global actors. This is where the work of the UN and its agencies, such as ITU or UNESCO,
becomes vital. The UN’s emphasis on ethical principles, human rights, and inclusivity ensures
that the global governance of AI is rooted in values that transcend borders. The success of these
and future initiatives will depend on the ability of the global community to work together, build
bridges between different regulatory approaches and ensure that AI governance frameworks
are not only compatible but also complementary.

Implementing AI governance frameworks is a global challenge that spans multiple sectors
as well as the entire AI supply chain. Striving for as much global consensus as possible is
important for effective governance, yet the complexity of national and regional regulations
or guidelines makes political solutions at the global level both challenging and crucial. AI
governance must address the borderless implications of the technology, requiring robust
international coordination. While governance frameworks must not stifle innovation, it is equally
important that competition between companies and countries does not undermine the integrity
of these frameworks. Practical steps toward achieving interoperability include the development
of AI Safety Institutes that are aligned to the same set of standards and principles as well as
conducting rigorous testing and sharing the outcomes of that testing to increase public trust,
ensuring that AI models meet international benchmarks for safety and ethics. These institutes
should collaborate globally to share insights and harmonize their approaches, thereby reducing
redundancy and enhancing the effectiveness of governance efforts.

21