Page 74 - AI Standards for Global Impact: From Governance to Action

AI Standards for Global Impact: From Governance to Action

• Support and encourage the development of conformity assessment frameworks
specifically targeting multimedia content, incorporating requirements related to AI risks,
misinformation, disinformation, and deepfakes.
• Consider a conformity assessment and/or certification scheme for multimedia content
authentication based on international standards, including relevant testing.

For technology developers and providers, policymakers could request that they consider the
following:

• Adopt a PDR framework based on internationally recognized standards to structure
responses to content authenticity challenges.
• Align with and monitor international standards and best practices to meet regulatory
requirements and future-proof innovation pipelines.
• Assign a standards liaison or champion within your organization to track updates, ensure
compliance, and guide the integration of emerging requirements.
• Consider the integration of strong cryptographic protocols, such as public key infrastructure
(PKI_, to enable secure multimedia authentication and content integrity.
• Leverage secure timestamping, tamper-evident hashes, and digital signatures to verify
content authenticity while preserving user privacy.

10�4 Fighting misinformation through fact-checking and deepfake
detection

Fact-checkers play an essential role in today’s information environment by helping to limit the
sharing of misinformation or disinformation. This session explored the role of fact-checkers
and the tools used in the process of verifying information and aimed to provide practical
guidelines for financial and social media companies on how to verify visual misinformation
and disinformation.

The session brought together stakeholders from diverse backgrounds. TikTok, WITNESS,
ElevenLabs, Ant Group, and Umanitek shared their views and insights on how they deal with
fact-checking and misinformation and the tools that they use.
WITNESS shared its global experience supporting frontline journalists and fact-checkers on
the detection of deceptive AI, particularly in election and conflict contexts. Examples were
highlighted with the Deepfakes Rapid Response Force and from within related trainings and
training materials. It was highlighted that there is a need to develop standards for multimedia
authenticity and WITNESS shared insight on its work in C2PA.
ElevenLabs shared current trends in deepfakes, the state of cooperation between governments
and the technology industry, and what the technology industry has already done and can do
in future. It was highlighted that there is a need for governments and the technology industry
to collaborate

TikTok highlighted its work to maintain platform integrity. Insights were provided about its
global fact-checking programme that includes more than 20 fact-checking partners, covering
more than 60 languages across more than 130 markets. Maintaining platform integrity is crucial
to providing a safe space for its users to enjoy authentic content, said TikTok. Alongside fact-
checking, TikTok uses a combination of advanced moderation technologies and teams of safety
experts. AI is being used to strengthen moderation efforts. In 2024, more than 80% of violent

62