On Thursday, December 3rd, President Trump signed an Executive Order (EO) on Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government. The EO lays out nine Principles to guide agencies (excluding the Department of Defense and Intelligence Community) in the development, acquisition, and use of these technologies — tasking the Office of Management and Budget (OMB) with carrying out the necessary policy guidance.
The EO identifies the following Principles for Use of AI in Government:
- Lawful and respectful of US values. Agencies shall ensure AI is used in a manner consistent with US law and addresses privacy, civil rights, and civil liberties.
- Purposeful and performance-driven. Agencies shall seek opportunities to use AI where the benefits significantly outweigh the risks, and the risks can be assessed and managed.
- Accurate, reliable, and effective. AI applications shall be consistent with the use cases for which the AI was trained — ensuring accuracy, reliability, and effectiveness.
- Safe, secure, and resilient. Agencies shall ensure the safety, security, and resiliency of AI, including when faced with systematic vulnerabilities, adversarial manipulation, and other malicious exploitation.
- Understandable. Operations and outcomes of AI applications must also be sufficiently understandable by subject matter experts, users, and others, as appropriate.
- Responsible and traceable. Human roles and responsibilities must be clearly defined, understood, and appropriately assigned for the development, acquisition, and use of AI, and relevant inputs and outputs of applications should be well documented and traceable.
- Regularly monitored. AI applications shall be regularly tested against these Principles, and mechanisms to deactivate applications that are inconsistent this order shall be maintained
- Transparent. Agencies shall be transparent in disclosing relevant information regarding use of AI to appropriate stakeholders, including the Congress and the public, to the extent practicable.
- Accountable. Agencies shall be accountable for implementing appropriate safeguards and shall monitor, audit, and document compliance, as well as provide training to agency staff.
OMB is tasked with implementing the above Principles through the necessary policy guidance. Within 180 days, OMB shall develop a roadmap of policies, including timelines for engaging the public. The EO also advises agencies to continue using voluntary consensus standards developed with industry and advises OMB to consider these standards in the development of new guidance.
Further, President Trump’s EO tasks the Federal Chief Information Officers Council (CIO Council) with identifying and making publicly available the criteria, format, and mechanisms for inventories of non-classified and non-sensitive AI use cases. As part of this process, agencies will identify and assess existing AI operations for any inconsistencies with this order and will implement plans to achieve consistency with AI applications or retire AI used in a manner that is inconsistent with this EO.
Finally, the EO provides guidance on interagency coordination and bodies by which agencies may participate, as well as guidance relating to AI implementation expertise. To achieve this latter point, the Presidential Innovation Fellows (PIF) program, in coordination with other agencies, will identify priority areas of expertise and establish an AI track to attract experts from industry and academia to work at government agencies.
Since the future of this EO is uncertain due to the transition of power at the White House, companies should engage with both the Biden-Harris Transition Team, as well as career agency officials, to inform a future iteration of this policy guidance under the next Administration.