Hello, Guest.!
BIS Outlines Rules Framework for Advanced AI Models, Chips Distribution
//

BIS Outlines Rules Framework for Advanced AI Models, Chips Distribution

2 mins read

The Bureau of Industry and Security has adopted three strategies to regulate the distribution of advanced computing chips and artificial intelligence models and prevent malicious actors’ access to advanced U.S. technologies in AI and semiconductors.

The first strategy calls for requiring authorizations for exports, reexports and in-country transfers of advanced computing chips. The strategy’s framework also defines certain exemptions on licenses and authorizations to ensure that risk-free commercial chip deals proceed and AI benefits are widely shared, BIS said Monday. 

The exceptions on advanced computing chips will cover a selection of U.S. allies and partners. Also exempted are supply chains in the chip export transfers for semiconductor development, production and storage, but excluding shipments to countries under U.S. arms embargo. In addition, the license exception will apply to low-volume chip transfers outside of the arms-embargoed countries.

Controls Based on AI Computational Weights 

The second strategy in the BIS framework requires export authorizations on AI model weights from 10^26 and above. The control also applies to certain computational model weights for AI developed abroad using advanced U.S. semiconductors or equipment. Exceptions to the rule include open-weight AI models deployed by entities based in the United States or in allied countries.

The third step in the BIS strategic framework calls for instituting security measures to protect the storage of advanced U.S. AI models and microchips overseas and mitigate the risk of the technologies’ diversion to U.S. adversaries.

Commerce Secretary Gina Raimondo noted that national security risk management also needs to consider AI evolution and U.S. allies’ desire to share in the technology’s benefits.

“This policy will help build a trusted technology ecosystem around the world and allow us to protect against the national security risks associated with AI, while ensuring controls do not stifle innovation or U.S. technological leadership,” she said.