OpenAI and Partners Establish International Network for AI Standards in 2024

In 2024, the International Network for Advanced AI Measurement, Evaluation, and Science will be established to address AI safety and measurement standards. This initiative, backed by organizations including OpenAI, the Center for AI Standards and Innovation (CAISI), and the Frontier Model Forum, aims to create international standards for safety and security practices in frontier AI development.

Sam Altman, co-founder of OpenAI, emphasized that the mission is to ensure artificial general intelligence benefits all of humanity. The network will focus on three main goals: enhancing AI safety, improving evaluation methods, and fostering collaboration among stakeholders.

The initiative will also involve contributions from the Appia Foundation and the Agentic AI Foundation, both of which are dedicated to developing AI specifications. OpenAI noted that international standards may be as crucial to pacing AI development as alignment research itself.

The establishment of this network follows a growing recognition of the need for structured oversight in AI, particularly after incidents involving Hugging Face that highlighted AI risks. Currently, nine countries have established AI safety institutes, reflecting a global effort to standardize practices in the field.

Automated AI research, as noted by OpenAI, can involve varying degrees of human supervision, indicating a shift towards more autonomous systems. This development is expected to influence how AI practitioners design and implement safety protocols in their projects moving forward.