New research shows that country-of-origin labels can obscure an AI model’s upstream dependencies, inherited behaviors and potential security risks.
The post Think You’ve Eliminated Chinese AI? Check the Model’s Lineage, Cisco Says appeared first on SecurityWeek.