Governance & Risk Management , Standards, Regulations & Compliance

OMB Issues First Governmentwide AI Risk Mitigation Rules

U.S. federal agencies have until December to implement a series of safeguards that aim to ensure the government is responsibly using artificial intelligence, the White House ordered Thursday.

See Also: Alleviating Compliance Pain Points in the Cloud Era

A memo from the Office of Management and Budget requires agencies to designate chief AI officers and establish AI governance boards to coordinate and govern the use of the emerging technology across the organization. Agencies should also add more detail to annual reports about which AI tools they use and release government-owned AI code.

"Agencies must improve their ability to use AI in ways that benefit the public and increase mission effectiveness, while recognizing the limitations and risks of AI," the memo says. "Agencies are encouraged to prioritize AI development and adoption for the public good and where the technology can be helpful in understanding and tackling large societal challenges."

The 34-page memorandum was released as the administration announced that federal agencies successfully completed all of the 150-day actions included in President Joe Biden's October executive order on AI. If an agency is unable to comply with the latest guidance, it is required to cease using the AI system - unless agency leadership can show that doing so would create an unacceptable disruption of critical operations.

The memo requires agencies to develop and publicly release a strategy for identifying and removing barriers to the responsible use of AI, including steps to achieve enterprisewide improvements in AI maturity. OMB said it will issue a request for information to help develop regulations around the federal procurement of AI tools and technologies.

Several agencies - including the departments of State, Defense, Veterans Affairs and Housing and Urban Development - have already established AI governance boards. The rest of the 24 largest federal agencies have until May 27 to form their own governance bodies.

The memo directs agencies to ensure their AI projects have access to adequate information technology infrastructure, "including high-performance computing infrastructure specialized for AI, training and interference." The White House also encourages agencies to ensure that AI developers have access to open-source libraries, software tools and monitoring capabilities to oversee federal AI applications.

The memo calls for a significant level of transparency into federal AI tools, requiring agencies to "proactively share their custom-developed code" for AI applications in active use via open-source software on a public repository, unless sharing of the code is restricted by law or regulations or would create contractual issues and national security risks.

"Agencies should prioritize sharing custom-developed code, such as commonly used packages or functions, that has the greatest potential for re-use by other agencies or the public," the memo reads.

The administration announced commitments to strengthen the federal cyber workforce, including hiring 100 AI professionals by summer 2024. The White House federal spending blueprint for the next federal fiscal year contains a $5 million proposal to expand the General Services Administration's governmentwide AI training program. More than 4,800 individuals across 78 federal agencies participated in the initiative last year, according to OMB.

OMB officials told reporters Wednesday the agency will now begin working on further action required under the AI executive order around federal procurement of emerging technologies due 180 days after the release of the memorandum.