Page 71 - AI Standards for Global Impact: From Governance to Action

AI Standards for Global Impact: From Governance to Action

The paper offers practical guidance and actionable recommendations, including a regulatory
options matrix designed to help policymakers and regulators determine what to regulate
(scope), how to regulate (voluntary or mandatory mechanisms), and to what extent (level of
effort). It also explores a range of supporting tools – such as standards, conformity assessment Part 2: Thematic AI
mechanisms, and enabling technologies – that can contribute to addressing the challenges
of misinformation and disinformation related to multimedia content. At the same time, it
emphasizes the importance of striking a balance that enables the positive and legitimate use of
either fully or partially synthetic multimedia for societal, governmental, and commercial benefit.

One of the major challenges faced by policymakers and regulators is that multimedia authenticity
in the case of generative AI is fundamentally a "Black Box."’. There is limited transparency
about how these models are developed and trained. The question that looms is how to enable
effective governance when the underlying operations are largely opaque. The main challenges
about how to ensure trustworthiness and interpretability of multimedia content without stifling
innovation intersects with broader concerns. These include how to align with emerging global
priorities, such as combatting misinformation, and how they can be shaped or influenced by
online safety regulations.

According to the paper, there is a recognition from multiple stakeholders that regulatory and
enforcement bodies cannot alone build trust in multimedia. All stakeholders need to work
together and find new forms of international collaboration and regulation, even perhaps
self-regulation. This needs to be coupled with corporate responsibility that fosters trust and
promotes human rights, media literacy, and ethics.

The paper proposes the adoption of Prevent-Detect-Respond (PDR) frameworks to build trust
in multimedia authenticity to address the above challenges. This three-pronged approach aims
to provide a scalable, flexible structure that balances regulatory intent with technical feasibility.
This framework mirrors existing approaches to privacy (e.g. GDPR and the California Consumer
Privacy Act) and cybersecurity (e.g. NIST cybersecurity framework and the Payment Card Industry
Data Security Standards). The strength of PDR lies in its simplicity and versatility.; it is widely
understood, adaptable throughout sectors, and conducive to regulatory alignment. In the
case of privacy, successful approaches emphasize prevention (privacy-by-design), detection
(breach notification and monitoring), and response (enforcement actions and mechanisms for
user redress).

59