Multi-Association Request for a NIST AI Framework

  • LETTER TO CONGRESS

Chairman José Serrano
House Subcommittee on Commerce,
Justice, Science and Related Agencies
H-310 The Capitol
Washington, DC 20515

Ranking Member Robert Aderholt
House Subcommittee on Commerce,
Justice, Science and Related Agencies
H-310 The Capitol
Washington, DC 20515

Dear Chairman Serrano and Ranking Member Aderholt:

On behalf of the undersigned organizations, we urge you to include language in the upcoming FY 21 CJS Appropriations Act to preserve the United States’ leading role in the development of Artificial Intelligence (“AI”). Specifically, we seek your support for report language – such as the proposals by Rep. Brenda Lawrence and Rep. Will Hurd – directing the National Institute for Standards and Technology (“NIST”) to initiate a process to begin developing an AI risk management framework.

US companies are on the leading edge of developing applications of AI that are reshaping the global economy and enhancing the competitiveness of American industries. Harnessing AI to transform raw data into actionable intelligence can help farmers increase their crop yields while minimizing waste, decrease wait times at hospitals while improving patient outcomes, and identify supply chain inefficiencies that can reduce shipping costs and greenhouse emissions. But as the benefits of AI are being realized, so too are the potential challenges. As AI is integrated into business processes that have meaningful impacts on people – affecting, for instance, access to credit, housing, or employment – significant risks can arise if systems fail unexpectedly, if they generate biased outcomes, or if they lack appropriate accountability safeguards.

Because the US is home to the world’s leading AI companies, the United States must stake out a leadership position in ensuring that the technology is being developed in a trustworthy and accountable manner. Of course, because the risks to the trustworthiness of AI are wide-ranging and application-specific, there are no one-size-fits-all technical or policy solutions. Instead, experts agree that a risk management approach that extends throughout an AI system’s lifecycle is necessary. Given its considerable experience developing lifecycle risk management frameworks for cybersecurity and privacy, NIST should play a leading role in the development of a similar framework for AI in close collaboration with stakeholders. This work would be in concert with NIST’s plan for prioritizing federal agency engagement in AI standards development as directed by the Executive Order on Maintaining American Leadership Artificial Intelligence (EO 13859). We recognize that standards and best practices will be key inputs to the long-term success of the risk management framework and urge NIST to continue that work.

By partnering with industry in the development of such a framework, NIST can help forge a shared conceptual foundation that will support the development of trustworthy AI. Given the quickly evolving nature of AI, an AI framework can serve as a helpful resource for organizations as they evaluate their own uses of AI and seek to identify the range of standards, best practices, governance approaches, and technical safeguards that can be implemented through a system’s lifecycle to mitigate potential risks.

A NIST voluntary framework-based consensus set of best practices would be pro-innovation, support U.S. leadership, be consistent with NIST’s ongoing engagement on AI industry consensus standards development, and align with U.S. support for the OECD AI principles as well as the draft Memorandum to Heads of Executive Departments and Agencies, “Guidance for Regulation of Artificial Intelligence Applications.” Importantly, a NIST AI framework would also send a powerful signal that would help inform US trading partners that are currently considering their own policy approaches for AI. NIST’s credibility in developing technically-informed frameworks for managing the risks associated with emerging technologies is unparalleled. A NIST-backed AI framework can therefore help shape the global debate around AI governance in ways that will inure to the benefit of the United States.

Thank you in advance for your consideration of this request.

Sincerely,

BSA | The Software Alliance
Chamber Technology Engagement Center (C_TEC)
Computing Technology Industry Association (CompTIA)
Consumer Technology Association (CTA)
Information Technology Industry Council (ITI)