The White House is set to issue a sweeping executive order aimed at securing the development of advanced artificial intelligence (AI) models. President Joe Biden will invoke executive powers over private industry to require developers of generative AI foundation models, with the potential to pose risks to national security, to notify the government during their training and share results of safety tests.
The order utilizes the Defense Production Act, a 1950 statute, and the National Institute of Standards and Technology will establish new standards for extensive red-team testing of foundational models. Additionally, the Department of Health and Human Services will establish a safety program to address AI-related healthcare practices.
White House deputy chief of staff Bruce Reed described the executive order as an unprecedented set of actions for AI safety, security, and trust, emphasizing its strength. The order further mandates the Department of Homeland Security to apply the National Institute of Standards and Technology (NIST) standards to critical infrastructure sectors and establish an AI Safety and Security Board.
A government cybersecurity program will be initiated to develop AI security tools, address vulnerabilities in AI technologies, and set new standards for the use of AI systems within federal agencies. In alignment with positioning the United States as a leader in AI research, the order also focuses on expanding grants across key sectors, such as healthcare and climate change, supported by the National AI Research Resource pilot program.
The U.S. is taking comprehensive measures to address AI risks, both nationally and internationally. The order builds upon previous efforts by the Biden administration to establish guidelines for the deployment of AI systems, combat algorithmic discrimination, and secure commitments from leading AI companies to meet rigorous safety standards, enhancing trust and safety in the AI industry. The U.S. aims to remain at the forefront of AI technology development while ensuring its responsible and secure application.
