On December 3, 2020, the President issued Executive Order 13960, Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government. The Executive Order signals yet again the Federal Government’s emphasis on the use of AI and the importance of ensuring that AI is used in a manner that does not violate Americans’ privacy, civil rights, and civil liberties, consistent with the previously issued Executive Order 13859 and Department of Defense policy announcements covered here, here, and here. The Order encourages agencies to continue to use AI and establishes principles for their use, at least for purposes other than national security and defense.

The principles established in the Executive Order are as follows:

1. Agencies must use AI in a manner that respects the Nation’s values and is consistent with the Constitution and other laws and regulations.
2. Agencies must assess and manage the risks of using AI and ensure that the benefits of using AI outweigh the risks.
3. Agencies must ensure that the application of AI is accurate, reliable, and effective.
4. Agencies must ensure that AI applications are safe, secure, and resilient.
5. Agencies must ensure that the outputs from AI applications are understandable.
6. Agencies must ensure that human responsibilities are clearly defined for AI applications, and that inputs for and outputs from AI applications are documented and traceable.
7. Agencies must ensure that AI applications are regularly monitored, including testing each application against these established principles.
8. Agencies must be transparent regarding their use of AI and provide relevant information to appropriate stakeholders like Congress and the public.
9. Finally, agencies shall be accountable for ensuring proper safeguards for the use and function of AI applications.

These principles are likely to play a role in future procurements and applications of AI, including (1) how agencies procure AI applications, (2) what information agencies will require contractors to provide regarding their AI applications, and (3) how agencies test and audit government contracts for AI applications or government contracts whose performance relies on the use of AI applications.

We will continue to monitor and report on these developments.