The Biden administration issues guidelines to federal agencies buying AI

That includes establishing teams to work together on informing future AI policy

The Biden administration has made ethics around artificial intelligence a priority: a year ago, it issued an Executive Order (EO) on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence, directing action to strengthen AI safety and security, while also protecting the privacy of Americans.

Now the government is out with new guidelines for agencies when it comes to buying AI, issuing a memo called the Advancing the Responsible Acquisition of Artificial Intelligence in Government, aka M-24-18 last week.

The memo details how agencies should appropriately manage risks and performance; promote a competitive marketplace; and implement structures to govern and manage their business processes related to acquiring AI.

"As the largest single buyer in the United States economy with over $750 billion in annual spend, the Federal Government’s procurement decisions can have far-reaching implications," the Office of Management and Budget (OMB) wrote in a press release.

"In 2023, the Federal Government purchased more than $100 billion in IT products and services alone, and the exercise of this purchasing power can accelerate technological advancements in critical areas, including AI."

M-24-18 builds on OMB M-24-10, issued in March 2024, which introduced government-wide binding requirements for agencies to strengthen governance, innovation, and risk management for use of AI.

The new memorandum requires that agency privacy officials and programs have involvement in AI acquisition processes, allowing them to identify and manage privacy risks and ensure compliance with law and policy, while calling for agencies to work with vendors so they can understand when AI is being acquired, and when that might trigger additional risk management requirements around rights and safety. The memo also promotes acquisition techniques that give agencies the ability to plan for and manage risk.

"The complex nature of how AI systems are built, trained, and deployed creates certain considerations and challenges for agency acquisition of AI.  For this reason, M-24-18 includes best practices and specific requirements for managing AI risk and performance, with additional requirements for acquiring AI use cases associated with rights-impacting and safety-impacting AI," wrote OMB. 

Agencies are now required to establish cross-functional teams that include officials with AI expertise and personnel from other relevant fields, such as acquisition, cybersecurity, privacy, and civil liberties, to inform strategic planning and acquisition of AI. The idea is for agencies to work together, sharing lessons learned, to inform future policy.

That means, for example, identifying and prioritizing AI investments that best serve an agency’s mission; developing the capacity to deploy any acquired AI; and promoting adoption of cross-functional best practices for the duration of use.

"As a part of the Biden-Harris Administration’s comprehensive strategy for responsible innovation, the guidance is designed to serve as a first step toward helping agencies and vendors grow together as the AI market continues to evolve – charting the course for ensuring that Federal acquisition of AI enables agencies to responsibly optimize the services they deliver for the American people," wrote OMB. 

(Image source: fedscoop.com)