Meta has unveiled a new risk policy framework to evaluate and mitigate risks posed by advanced AI models, categorizing them into high and critical risk groups. The Frontier AI Framework outlines scenarios like biological weapon proliferation and economic impact via fraud. Critical-risk models may face halted development and restricted access, while high-risk models will see limited access and mitigations. Meta's approach involves multidisciplinary engagement and aims to enhance AI safety through robust evaluations.
Meta has unveiled a new risk policy framework to evaluate and mitigate risks posed by advanced AI models, categorizing them into high and critical risk groups. The Frontier AI Framework outlines scenarios like biological weapon proliferation and economic impact via fraud. Critical-risk models may face halted development and restricted access, while high-risk models will see limited access and mitigations. Meta's approach involves multidisciplinary engagement and aims to enhance AI safety through robust evaluations.