Response Letter to OMB’s ‘Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence’ Draft Memorandum

On October 30, President Biden signed the landmark Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (fact sheet), followed shortly by draft guidance from the Office of Management and Budget (fact sheet). These build on the previously released Blueprint for an AI Bill of Rights and NIST AI Risk Management Framework that provided governments with guidance on the responsible and ethical use of AI.

To improve the guidance for federal agencies’ governance of AI with a focus on improving the digital delivery of government programs, the Beeck Center for Social Impact + Innovation at Georgetown University compiled 10 recommendations for OMB’s consideration in response to their request for comment via regulations.gov. These recommendations are in response to the Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence Draft Memorandum. They include:

1. Clarifying the overlapping responsibilities and points of coordination between appointed chief AI officers and existing executives overseeing IT, data, or cybersecurity. 
2. Intentionally and strategically building AI governance from the existing data-governance infrastructure and aligning it with data privacy, risk assessment, open data, and other data-management practices that form the foundation of responsible AI innovation in government. 
3. Strengthening measures to ensure agencies actually disclose AI uses in their inventories. 
4. Providing guidance for appropriate oversight of removal of barriers to the responsible use of AI.
5. Clarifying how waivers to minimum practice requirements are established and potentially strengthening OMB’s oversight of the waiver process.
6. Exploring opportunities for improving business operations and delivery of government programs through generative AI.
7. Requiring cross-agency collaboration in developing guidance for the use of generative AI in public benefits.
8. Providing guidance for how other automated, but non-generative technologies, such as robotic process automation, can work separately and in conjunction with generative AI to improve business operations. 
9. Developing and publishing an “oversight guide” for reviewing agency acquisition activities.
10. Defining “adequate performance,” including by providing suggested documentation that vendors must submit to support performance measurement and evaluation for AI contracts. 

Check our more detailed recommendations and resources here.