United States of America: Introduced Safe and Secure Innovation for Frontier Artificial Intelligence Systems Act (SB 1047) including testing requirements

On 7 February 2024, the Safe and Secure Innovation for Frontier Artificial Intelligence (AI) Systems Act (SB 1047) was introduced in the California Senate. The Act authorises Developers of covered systems to determine whether a positive safety determination can be made. This determination, excluding derivative models, is defined as whether "a developer can reasonably exclude the possibility that a covered model has a hazardous capability or may come close to possessing a hazardous capability when accounting for a reasonable margin for safety and the possibility of posttraining modifications." Developers must comply with various requirements, including the ability to promptly initiate a complete shutdown of the covered model. Further, the Act mandates that developers conduct performance benchmarking, and submit certifications to the Frontier Model Division to ensure that AI models are developed with safety and security as priorities. The Act requires detailed safety protocols, regular audits, and compliance with transparency measures to mitigate risks associated with hazardous capabilities and promote accountability.