To embed, copy and paste the code into your website or blog:
The National Defense Authorization Act for the Fiscal Year 2021 (2021 NDAA)—which was passed over a Presidential veto on January 1—represents a massive step forward for American AI policy in areas far beyond national defense. It incorporates a number of AI legislative proposals that will reshape the government’s approach over the next two years, as part of a broader emphasis on promoting emerging technologies.
Among its many elements, the 2021 NDAA (1) expands National Institute of Standards and Technology’s (NIST) AI responsibilities, including directing it to establish a voluntary risk management framework—in consultation with industry—that will identify and provide standards and best practices for assessing the trustworthiness of AI and mitigating risks from AI systems; (2) launches the National Artificial Intelligence Initiative, setting up a federal bureaucracy designed to deal both with agencies and outside stakeholders, as well as advise on key issue issues of AI implementation like bias and fairness; (3) gives the Department of Defense (DoD) specific authority to procure AI while requiring an assessment meant to promote acquisition of AI that is “ethically and responsibly developed.” All of these initiatives will have ripple effects on private sector development, testing, and deployment of AI systems—and heavily influence regulatory expectations on issues like AI bias, accuracy, and security.