AI giants team up to form "frontier AI" models
A new industry organization created to assure "safe and responsible development" of so-called "frontier AI" models has been formed by four of the leading AI players, OpenAI, Google, Microsoft, and Anthropic.
OpenAI, Microsoft, Google, and Anthropic, developers of popular AI tools, have introduced the Frontier Model Forum in response to growing calls for regulatory oversight. This alliance leans on the knowledge of its member businesses to create technical evaluations and benchmarks and to advance best practices and standards.
The Frontier AI Model Forum currently only has four members, but it claims to be open to newcomers. Organizations that meet the criteria for qualification should be working on cutting-edge AI models and exhibiting a "strong commitment to frontier model safety."
Details about the Frontier AI Forum
Frontier AI, which is cutting-edge AI and machine learning models deemed dangerous enough to pose "serious risks to public safety," is at the heart of the forum's goal.
The Forum's main goals will be to promote information exchange between businesses and governments, identify best practices for creating and deploying AI models, and enhance research into AI safety. Testing and comparing various AI models as well as publishing the benchmarks in a public library for every significant industry participant to grasp the potential hazards and take appropriate action will all be included in the AI safety research.
OpenAI couldn't "classify" enough
In order to set strategies and priorities, the Frontier Model Forum will create an advisory board made up of participants with various backgrounds and viewpoints. Additionally, founding businesses will create institutional structures that will enable all stakeholders to work together to reach a consensus on matters pertaining to the advancement of AI.
The Forum will also use the information and materials that researchers, civil society organizations, and other parties have already offered. In order to effectively implement the standards established by existing AI initiatives like Partnership on AI and MLCommons, it will observe and support them.
Advertisement
